Mac: finish Moltbot rename

This commit is contained in:
Shadow
2026-01-27 14:12:17 -06:00
parent 3fe4b2595a
commit cc72498b46
77 changed files with 18956 additions and 44 deletions

View File

@@ -13,7 +13,7 @@ android {
sourceSets { sourceSets {
getByName("main") { getByName("main") {
assets.srcDir(file("../../shared/ClawdbotKit/Sources/ClawdbotKit/Resources")) assets.srcDir(file("../../shared/MoltbotKit/Sources/MoltbotKit/Resources"))
} }
} }

View File

@@ -15,7 +15,7 @@ open Clawdbot.xcodeproj
``` ```
## Shared packages ## Shared packages
- `../shared/ClawdbotKit` — shared types/constants used by iOS (and later macOS bridge + gateway routing). - `../shared/MoltbotKit` — shared types/constants used by iOS (and later macOS bridge + gateway routing).
## fastlane ## fastlane
```bash ```bash

View File

@@ -24,37 +24,37 @@ Sources/Status/VoiceWakeToast.swift
Sources/Voice/VoiceTab.swift Sources/Voice/VoiceTab.swift
Sources/Voice/VoiceWakeManager.swift Sources/Voice/VoiceWakeManager.swift
Sources/Voice/VoiceWakePreferences.swift Sources/Voice/VoiceWakePreferences.swift
../shared/ClawdbotKit/Sources/ClawdbotChatUI/ChatComposer.swift ../shared/MoltbotKit/Sources/MoltbotChatUI/ChatComposer.swift
../shared/ClawdbotKit/Sources/ClawdbotChatUI/ChatMarkdownRenderer.swift ../shared/MoltbotKit/Sources/MoltbotChatUI/ChatMarkdownRenderer.swift
../shared/ClawdbotKit/Sources/ClawdbotChatUI/ChatMarkdownPreprocessor.swift ../shared/MoltbotKit/Sources/MoltbotChatUI/ChatMarkdownPreprocessor.swift
../shared/ClawdbotKit/Sources/ClawdbotChatUI/ChatMessageViews.swift ../shared/MoltbotKit/Sources/MoltbotChatUI/ChatMessageViews.swift
../shared/ClawdbotKit/Sources/ClawdbotChatUI/ChatModels.swift ../shared/MoltbotKit/Sources/MoltbotChatUI/ChatModels.swift
../shared/ClawdbotKit/Sources/ClawdbotChatUI/ChatPayloadDecoding.swift ../shared/MoltbotKit/Sources/MoltbotChatUI/ChatPayloadDecoding.swift
../shared/ClawdbotKit/Sources/ClawdbotChatUI/ChatSessions.swift ../shared/MoltbotKit/Sources/MoltbotChatUI/ChatSessions.swift
../shared/ClawdbotKit/Sources/ClawdbotChatUI/ChatSheets.swift ../shared/MoltbotKit/Sources/MoltbotChatUI/ChatSheets.swift
../shared/ClawdbotKit/Sources/ClawdbotChatUI/ChatTheme.swift ../shared/MoltbotKit/Sources/MoltbotChatUI/ChatTheme.swift
../shared/ClawdbotKit/Sources/ClawdbotChatUI/ChatTransport.swift ../shared/MoltbotKit/Sources/MoltbotChatUI/ChatTransport.swift
../shared/ClawdbotKit/Sources/ClawdbotChatUI/ChatView.swift ../shared/MoltbotKit/Sources/MoltbotChatUI/ChatView.swift
../shared/ClawdbotKit/Sources/ClawdbotChatUI/ChatViewModel.swift ../shared/MoltbotKit/Sources/MoltbotChatUI/ChatViewModel.swift
../shared/ClawdbotKit/Sources/ClawdbotKit/AnyCodable.swift ../shared/MoltbotKit/Sources/MoltbotKit/AnyCodable.swift
../shared/ClawdbotKit/Sources/ClawdbotKit/BonjourEscapes.swift ../shared/MoltbotKit/Sources/MoltbotKit/BonjourEscapes.swift
../shared/ClawdbotKit/Sources/ClawdbotKit/BonjourTypes.swift ../shared/MoltbotKit/Sources/MoltbotKit/BonjourTypes.swift
../shared/ClawdbotKit/Sources/ClawdbotKit/BridgeFrames.swift ../shared/MoltbotKit/Sources/MoltbotKit/BridgeFrames.swift
../shared/ClawdbotKit/Sources/ClawdbotKit/CameraCommands.swift ../shared/MoltbotKit/Sources/MoltbotKit/CameraCommands.swift
../shared/ClawdbotKit/Sources/ClawdbotKit/CanvasA2UIAction.swift ../shared/MoltbotKit/Sources/MoltbotKit/CanvasA2UIAction.swift
../shared/ClawdbotKit/Sources/ClawdbotKit/CanvasA2UICommands.swift ../shared/MoltbotKit/Sources/MoltbotKit/CanvasA2UICommands.swift
../shared/ClawdbotKit/Sources/ClawdbotKit/CanvasA2UIJSONL.swift ../shared/MoltbotKit/Sources/MoltbotKit/CanvasA2UIJSONL.swift
../shared/ClawdbotKit/Sources/ClawdbotKit/CanvasCommandParams.swift ../shared/MoltbotKit/Sources/MoltbotKit/CanvasCommandParams.swift
../shared/ClawdbotKit/Sources/ClawdbotKit/CanvasCommands.swift ../shared/MoltbotKit/Sources/MoltbotKit/CanvasCommands.swift
../shared/ClawdbotKit/Sources/ClawdbotKit/Capabilities.swift ../shared/MoltbotKit/Sources/MoltbotKit/Capabilities.swift
../shared/ClawdbotKit/Sources/ClawdbotKit/ClawdbotKitResources.swift ../shared/MoltbotKit/Sources/MoltbotKit/ClawdbotKitResources.swift
../shared/ClawdbotKit/Sources/ClawdbotKit/DeepLinks.swift ../shared/MoltbotKit/Sources/MoltbotKit/DeepLinks.swift
../shared/ClawdbotKit/Sources/ClawdbotKit/JPEGTranscoder.swift ../shared/MoltbotKit/Sources/MoltbotKit/JPEGTranscoder.swift
../shared/ClawdbotKit/Sources/ClawdbotKit/NodeError.swift ../shared/MoltbotKit/Sources/MoltbotKit/NodeError.swift
../shared/ClawdbotKit/Sources/ClawdbotKit/ScreenCommands.swift ../shared/MoltbotKit/Sources/MoltbotKit/ScreenCommands.swift
../shared/ClawdbotKit/Sources/ClawdbotKit/StoragePaths.swift ../shared/MoltbotKit/Sources/MoltbotKit/StoragePaths.swift
../shared/ClawdbotKit/Sources/ClawdbotKit/SystemCommands.swift ../shared/MoltbotKit/Sources/MoltbotKit/SystemCommands.swift
../shared/ClawdbotKit/Sources/ClawdbotKit/TalkDirective.swift ../shared/MoltbotKit/Sources/MoltbotKit/TalkDirective.swift
../../Swabble/Sources/SwabbleKit/WakeWordGate.swift ../../Swabble/Sources/SwabbleKit/WakeWordGate.swift
Sources/Voice/TalkModeManager.swift Sources/Voice/TalkModeManager.swift
Sources/Voice/TalkOrbOverlay.swift Sources/Voice/TalkOrbOverlay.swift

View File

@@ -11,7 +11,7 @@ settings:
packages: packages:
MoltbotKit: MoltbotKit:
path: ../shared/ClawdbotKit path: ../shared/MoltbotKit
Swabble: Swabble:
path: ../../Swabble path: ../../Swabble

View File

@@ -20,7 +20,7 @@ let package = Package(
.package(url: "https://github.com/apple/swift-log.git", from: "1.8.0"), .package(url: "https://github.com/apple/swift-log.git", from: "1.8.0"),
.package(url: "https://github.com/sparkle-project/Sparkle", from: "2.8.1"), .package(url: "https://github.com/sparkle-project/Sparkle", from: "2.8.1"),
.package(url: "https://github.com/steipete/Peekaboo.git", branch: "main"), .package(url: "https://github.com/steipete/Peekaboo.git", branch: "main"),
.package(path: "../shared/ClawdbotKit"), .package(path: "../shared/MoltbotKit"),
.package(path: "../../Swabble"), .package(path: "../../Swabble"),
], ],
targets: [ targets: [

View File

@@ -1,4 +1,4 @@
# Clawdbot macOS app (dev + signing) # Moltbot macOS app (dev + signing)
## Quick dev run ## Quick dev run
@@ -20,7 +20,7 @@ scripts/restart-mac.sh --sign # force code signing (requires cert)
scripts/package-mac-app.sh scripts/package-mac-app.sh
``` ```
Creates `dist/Clawdbot.app` and signs it via `scripts/codesign-mac-app.sh`. Creates `dist/Moltbot.app` and signs it via `scripts/codesign-mac-app.sh`.
## Signing behavior ## Signing behavior

View File

@@ -0,0 +1,340 @@
import Foundation
import OSLog
enum AgentWorkspace {
private static let logger = Logger(subsystem: "bot.molt", category: "workspace")
static let agentsFilename = "AGENTS.md"
static let soulFilename = "SOUL.md"
static let identityFilename = "IDENTITY.md"
static let userFilename = "USER.md"
static let bootstrapFilename = "BOOTSTRAP.md"
private static let templateDirname = "templates"
private static let ignoredEntries: Set<String> = [".DS_Store", ".git", ".gitignore"]
private static let templateEntries: Set<String> = [
AgentWorkspace.agentsFilename,
AgentWorkspace.soulFilename,
AgentWorkspace.identityFilename,
AgentWorkspace.userFilename,
AgentWorkspace.bootstrapFilename,
]
enum BootstrapSafety: Equatable {
case safe
case unsafe(reason: String)
}
static func displayPath(for url: URL) -> String {
let home = FileManager().homeDirectoryForCurrentUser.path
let path = url.path
if path == home { return "~" }
if path.hasPrefix(home + "/") {
return "~/" + String(path.dropFirst(home.count + 1))
}
return path
}
static func resolveWorkspaceURL(from userInput: String?) -> URL {
let trimmed = userInput?.trimmingCharacters(in: .whitespacesAndNewlines) ?? ""
if trimmed.isEmpty { return MoltbotConfigFile.defaultWorkspaceURL() }
let expanded = (trimmed as NSString).expandingTildeInPath
return URL(fileURLWithPath: expanded, isDirectory: true)
}
static func agentsURL(workspaceURL: URL) -> URL {
workspaceURL.appendingPathComponent(self.agentsFilename)
}
static func workspaceEntries(workspaceURL: URL) throws -> [String] {
let contents = try FileManager().contentsOfDirectory(atPath: workspaceURL.path)
return contents.filter { !self.ignoredEntries.contains($0) }
}
static func isWorkspaceEmpty(workspaceURL: URL) -> Bool {
let fm = FileManager()
var isDir: ObjCBool = false
if !fm.fileExists(atPath: workspaceURL.path, isDirectory: &isDir) {
return true
}
guard isDir.boolValue else { return false }
guard let entries = try? self.workspaceEntries(workspaceURL: workspaceURL) else { return false }
return entries.isEmpty
}
static func isTemplateOnlyWorkspace(workspaceURL: URL) -> Bool {
guard let entries = try? self.workspaceEntries(workspaceURL: workspaceURL) else { return false }
guard !entries.isEmpty else { return true }
return Set(entries).isSubset(of: self.templateEntries)
}
static func bootstrapSafety(for workspaceURL: URL) -> BootstrapSafety {
let fm = FileManager()
var isDir: ObjCBool = false
if !fm.fileExists(atPath: workspaceURL.path, isDirectory: &isDir) {
return .safe
}
if !isDir.boolValue {
return .unsafe(reason: "Workspace path points to a file.")
}
let agentsURL = self.agentsURL(workspaceURL: workspaceURL)
if fm.fileExists(atPath: agentsURL.path) {
return .safe
}
do {
let entries = try self.workspaceEntries(workspaceURL: workspaceURL)
return entries.isEmpty
? .safe
: .unsafe(reason: "Folder isn't empty. Choose a new folder or add AGENTS.md first.")
} catch {
return .unsafe(reason: "Couldn't inspect the workspace folder.")
}
}
static func bootstrap(workspaceURL: URL) throws -> URL {
let shouldSeedBootstrap = self.isWorkspaceEmpty(workspaceURL: workspaceURL)
try FileManager().createDirectory(at: workspaceURL, withIntermediateDirectories: true)
let agentsURL = self.agentsURL(workspaceURL: workspaceURL)
if !FileManager().fileExists(atPath: agentsURL.path) {
try self.defaultTemplate().write(to: agentsURL, atomically: true, encoding: .utf8)
self.logger.info("Created AGENTS.md at \(agentsURL.path, privacy: .public)")
}
let soulURL = workspaceURL.appendingPathComponent(self.soulFilename)
if !FileManager().fileExists(atPath: soulURL.path) {
try self.defaultSoulTemplate().write(to: soulURL, atomically: true, encoding: .utf8)
self.logger.info("Created SOUL.md at \(soulURL.path, privacy: .public)")
}
let identityURL = workspaceURL.appendingPathComponent(self.identityFilename)
if !FileManager().fileExists(atPath: identityURL.path) {
try self.defaultIdentityTemplate().write(to: identityURL, atomically: true, encoding: .utf8)
self.logger.info("Created IDENTITY.md at \(identityURL.path, privacy: .public)")
}
let userURL = workspaceURL.appendingPathComponent(self.userFilename)
if !FileManager().fileExists(atPath: userURL.path) {
try self.defaultUserTemplate().write(to: userURL, atomically: true, encoding: .utf8)
self.logger.info("Created USER.md at \(userURL.path, privacy: .public)")
}
let bootstrapURL = workspaceURL.appendingPathComponent(self.bootstrapFilename)
if shouldSeedBootstrap, !FileManager().fileExists(atPath: bootstrapURL.path) {
try self.defaultBootstrapTemplate().write(to: bootstrapURL, atomically: true, encoding: .utf8)
self.logger.info("Created BOOTSTRAP.md at \(bootstrapURL.path, privacy: .public)")
}
return agentsURL
}
static func needsBootstrap(workspaceURL: URL) -> Bool {
let fm = FileManager()
var isDir: ObjCBool = false
if !fm.fileExists(atPath: workspaceURL.path, isDirectory: &isDir) {
return true
}
guard isDir.boolValue else { return true }
if self.hasIdentity(workspaceURL: workspaceURL) {
return false
}
let bootstrapURL = workspaceURL.appendingPathComponent(self.bootstrapFilename)
guard fm.fileExists(atPath: bootstrapURL.path) else { return false }
return self.isTemplateOnlyWorkspace(workspaceURL: workspaceURL)
}
static func hasIdentity(workspaceURL: URL) -> Bool {
let identityURL = workspaceURL.appendingPathComponent(self.identityFilename)
guard let contents = try? String(contentsOf: identityURL, encoding: .utf8) else { return false }
return self.identityLinesHaveValues(contents)
}
private static func identityLinesHaveValues(_ content: String) -> Bool {
for line in content.split(separator: "\n") {
let trimmed = line.trimmingCharacters(in: .whitespacesAndNewlines)
guard trimmed.hasPrefix("-"), let colon = trimmed.firstIndex(of: ":") else { continue }
let value = trimmed[trimmed.index(after: colon)...].trimmingCharacters(in: .whitespacesAndNewlines)
if !value.isEmpty {
return true
}
}
return false
}
static func defaultTemplate() -> String {
let fallback = """
# AGENTS.md - Moltbot Workspace
This folder is the assistant's working directory.
## First run (one-time)
- If BOOTSTRAP.md exists, follow its ritual and delete it once complete.
- Your agent identity lives in IDENTITY.md.
- Your profile lives in USER.md.
## Backup tip (recommended)
If you treat this workspace as the agent's "memory", make it a git repo (ideally private) so identity
and notes are backed up.
```bash
git init
git add AGENTS.md
git commit -m "Add agent workspace"
```
## Safety defaults
- Don't exfiltrate secrets or private data.
- Don't run destructive commands unless explicitly asked.
- Be concise in chat; write longer output to files in this workspace.
## Daily memory (recommended)
- Keep a short daily log at memory/YYYY-MM-DD.md (create memory/ if needed).
- On session start, read today + yesterday if present.
- Capture durable facts, preferences, and decisions; avoid secrets.
## Customize
- Add your preferred style, rules, and "memory" here.
"""
return self.loadTemplate(named: self.agentsFilename, fallback: fallback)
}
static func defaultSoulTemplate() -> String {
let fallback = """
# SOUL.md - Persona & Boundaries
Describe who the assistant is, tone, and boundaries.
- Keep replies concise and direct.
- Ask clarifying questions when needed.
- Never send streaming/partial replies to external messaging surfaces.
"""
return self.loadTemplate(named: self.soulFilename, fallback: fallback)
}
static func defaultIdentityTemplate() -> String {
let fallback = """
# IDENTITY.md - Agent Identity
- Name:
- Creature:
- Vibe:
- Emoji:
"""
return self.loadTemplate(named: self.identityFilename, fallback: fallback)
}
static func defaultUserTemplate() -> String {
let fallback = """
# USER.md - User Profile
- Name:
- Preferred address:
- Pronouns (optional):
- Timezone (optional):
- Notes:
"""
return self.loadTemplate(named: self.userFilename, fallback: fallback)
}
static func defaultBootstrapTemplate() -> String {
let fallback = """
# BOOTSTRAP.md - First Run Ritual (delete after)
Hello. I was just born.
## Your mission
Start a short, playful conversation and learn:
- Who am I?
- What am I?
- Who are you?
- How should I call you?
## How to ask (cute + helpful)
Say:
"Hello! I was just born. Who am I? What am I? Who are you? How should I call you?"
Then offer suggestions:
- 3-5 name ideas.
- 3-5 creature/vibe combos.
- 5 emoji ideas.
## Write these files
After the user chooses, update:
1) IDENTITY.md
- Name
- Creature
- Vibe
- Emoji
2) USER.md
- Name
- Preferred address
- Pronouns (optional)
- Timezone (optional)
- Notes
3) ~/.clawdbot/moltbot.json
Set identity.name, identity.theme, identity.emoji to match IDENTITY.md.
## Cleanup
Delete BOOTSTRAP.md once this is complete.
"""
return self.loadTemplate(named: self.bootstrapFilename, fallback: fallback)
}
private static func loadTemplate(named: String, fallback: String) -> String {
for url in self.templateURLs(named: named) {
if let content = try? String(contentsOf: url, encoding: .utf8) {
let stripped = self.stripFrontMatter(content)
if !stripped.trimmingCharacters(in: .whitespacesAndNewlines).isEmpty {
return stripped
}
}
}
return fallback
}
private static func templateURLs(named: String) -> [URL] {
var urls: [URL] = []
if let resource = Bundle.main.url(
forResource: named.replacingOccurrences(of: ".md", with: ""),
withExtension: "md",
subdirectory: self.templateDirname)
{
urls.append(resource)
}
if let resource = Bundle.main.url(
forResource: named,
withExtension: nil,
subdirectory: self.templateDirname)
{
urls.append(resource)
}
if let dev = self.devTemplateURL(named: named) {
urls.append(dev)
}
let cwd = URL(fileURLWithPath: FileManager().currentDirectoryPath)
urls.append(cwd.appendingPathComponent("docs")
.appendingPathComponent(self.templateDirname)
.appendingPathComponent(named))
return urls
}
private static func devTemplateURL(named: String) -> URL? {
let sourceURL = URL(fileURLWithPath: #filePath)
let repoRoot = sourceURL
.deletingLastPathComponent()
.deletingLastPathComponent()
.deletingLastPathComponent()
.deletingLastPathComponent()
.deletingLastPathComponent()
return repoRoot.appendingPathComponent("docs")
.appendingPathComponent(self.templateDirname)
.appendingPathComponent(named)
}
private static func stripFrontMatter(_ content: String) -> String {
guard content.hasPrefix("---") else { return content }
let start = content.index(content.startIndex, offsetBy: 3)
guard let range = content.range(of: "\n---", range: start..<content.endIndex) else {
return content
}
let remainder = content[range.upperBound...]
let trimmed = remainder.trimmingCharacters(in: .whitespacesAndNewlines)
return trimmed + "\n"
}
// Identity is written by the agent during the bootstrap ritual.
}

View File

@@ -0,0 +1,384 @@
import CryptoKit
import Foundation
import OSLog
import Security
struct AnthropicOAuthCredentials: Codable {
let type: String
let refresh: String
let access: String
let expires: Int64
}
enum AnthropicAuthMode: Equatable {
case oauthFile
case oauthEnv
case apiKeyEnv
case missing
var shortLabel: String {
switch self {
case .oauthFile: "OAuth (Moltbot token file)"
case .oauthEnv: "OAuth (env var)"
case .apiKeyEnv: "API key (env var)"
case .missing: "Missing credentials"
}
}
var isConfigured: Bool {
switch self {
case .missing: false
case .oauthFile, .oauthEnv, .apiKeyEnv: true
}
}
}
enum AnthropicAuthResolver {
static func resolve(
environment: [String: String] = ProcessInfo.processInfo.environment,
oauthStatus: MoltbotOAuthStore.AnthropicOAuthStatus = MoltbotOAuthStore
.anthropicOAuthStatus()) -> AnthropicAuthMode
{
if oauthStatus.isConnected { return .oauthFile }
if let token = environment["ANTHROPIC_OAUTH_TOKEN"]?.trimmingCharacters(in: .whitespacesAndNewlines),
!token.isEmpty
{
return .oauthEnv
}
if let key = environment["ANTHROPIC_API_KEY"]?.trimmingCharacters(in: .whitespacesAndNewlines),
!key.isEmpty
{
return .apiKeyEnv
}
return .missing
}
}
enum AnthropicOAuth {
private static let logger = Logger(subsystem: "bot.molt", category: "anthropic-oauth")
private static let clientId = "9d1c250a-e61b-44d9-88ed-5944d1962f5e"
private static let authorizeURL = URL(string: "https://claude.ai/oauth/authorize")!
private static let tokenURL = URL(string: "https://console.anthropic.com/v1/oauth/token")!
private static let redirectURI = "https://console.anthropic.com/oauth/code/callback"
private static let scopes = "org:create_api_key user:profile user:inference"
struct PKCE {
let verifier: String
let challenge: String
}
static func generatePKCE() throws -> PKCE {
var bytes = [UInt8](repeating: 0, count: 32)
let status = SecRandomCopyBytes(kSecRandomDefault, bytes.count, &bytes)
guard status == errSecSuccess else {
throw NSError(domain: NSOSStatusErrorDomain, code: Int(status))
}
let verifier = Data(bytes).base64URLEncodedString()
let hash = SHA256.hash(data: Data(verifier.utf8))
let challenge = Data(hash).base64URLEncodedString()
return PKCE(verifier: verifier, challenge: challenge)
}
static func buildAuthorizeURL(pkce: PKCE) -> URL {
var components = URLComponents(url: self.authorizeURL, resolvingAgainstBaseURL: false)!
components.queryItems = [
URLQueryItem(name: "code", value: "true"),
URLQueryItem(name: "client_id", value: self.clientId),
URLQueryItem(name: "response_type", value: "code"),
URLQueryItem(name: "redirect_uri", value: self.redirectURI),
URLQueryItem(name: "scope", value: self.scopes),
URLQueryItem(name: "code_challenge", value: pkce.challenge),
URLQueryItem(name: "code_challenge_method", value: "S256"),
// Match legacy flow: state is the verifier.
URLQueryItem(name: "state", value: pkce.verifier),
]
return components.url!
}
static func exchangeCode(
code: String,
state: String,
verifier: String) async throws -> AnthropicOAuthCredentials
{
let payload: [String: Any] = [
"grant_type": "authorization_code",
"client_id": self.clientId,
"code": code,
"state": state,
"redirect_uri": self.redirectURI,
"code_verifier": verifier,
]
let body = try JSONSerialization.data(withJSONObject: payload, options: [])
var request = URLRequest(url: self.tokenURL)
request.httpMethod = "POST"
request.httpBody = body
request.setValue("application/json", forHTTPHeaderField: "Content-Type")
let (data, response) = try await URLSession.shared.data(for: request)
guard let http = response as? HTTPURLResponse else {
throw URLError(.badServerResponse)
}
guard (200..<300).contains(http.statusCode) else {
let text = String(data: data, encoding: .utf8) ?? "<non-utf8>"
throw NSError(
domain: "AnthropicOAuth",
code: http.statusCode,
userInfo: [NSLocalizedDescriptionKey: "Token exchange failed: \(text)"])
}
let decoded = try JSONSerialization.jsonObject(with: data) as? [String: Any]
let access = decoded?["access_token"] as? String
let refresh = decoded?["refresh_token"] as? String
let expiresIn = decoded?["expires_in"] as? Double
guard let access, let refresh, let expiresIn else {
throw NSError(domain: "AnthropicOAuth", code: 0, userInfo: [
NSLocalizedDescriptionKey: "Unexpected token response.",
])
}
// Match legacy flow: expiresAt = now + expires_in - 5 minutes.
let expiresAtMs = Int64(Date().timeIntervalSince1970 * 1000)
+ Int64(expiresIn * 1000)
- Int64(5 * 60 * 1000)
self.logger.info("Anthropic OAuth exchange ok; expiresAtMs=\(expiresAtMs, privacy: .public)")
return AnthropicOAuthCredentials(type: "oauth", refresh: refresh, access: access, expires: expiresAtMs)
}
static func refresh(refreshToken: String) async throws -> AnthropicOAuthCredentials {
let payload: [String: Any] = [
"grant_type": "refresh_token",
"client_id": self.clientId,
"refresh_token": refreshToken,
]
let body = try JSONSerialization.data(withJSONObject: payload, options: [])
var request = URLRequest(url: self.tokenURL)
request.httpMethod = "POST"
request.httpBody = body
request.setValue("application/json", forHTTPHeaderField: "Content-Type")
let (data, response) = try await URLSession.shared.data(for: request)
guard let http = response as? HTTPURLResponse else {
throw URLError(.badServerResponse)
}
guard (200..<300).contains(http.statusCode) else {
let text = String(data: data, encoding: .utf8) ?? "<non-utf8>"
throw NSError(
domain: "AnthropicOAuth",
code: http.statusCode,
userInfo: [NSLocalizedDescriptionKey: "Token refresh failed: \(text)"])
}
let decoded = try JSONSerialization.jsonObject(with: data) as? [String: Any]
let access = decoded?["access_token"] as? String
let refresh = (decoded?["refresh_token"] as? String) ?? refreshToken
let expiresIn = decoded?["expires_in"] as? Double
guard let access, let expiresIn else {
throw NSError(domain: "AnthropicOAuth", code: 0, userInfo: [
NSLocalizedDescriptionKey: "Unexpected token response.",
])
}
let expiresAtMs = Int64(Date().timeIntervalSince1970 * 1000)
+ Int64(expiresIn * 1000)
- Int64(5 * 60 * 1000)
self.logger.info("Anthropic OAuth refresh ok; expiresAtMs=\(expiresAtMs, privacy: .public)")
return AnthropicOAuthCredentials(type: "oauth", refresh: refresh, access: access, expires: expiresAtMs)
}
}
enum MoltbotOAuthStore {
static let oauthFilename = "oauth.json"
private static let providerKey = "anthropic"
private static let moltbotOAuthDirEnv = "CLAWDBOT_OAUTH_DIR"
private static let legacyPiDirEnv = "PI_CODING_AGENT_DIR"
enum AnthropicOAuthStatus: Equatable {
case missingFile
case unreadableFile
case invalidJSON
case missingProviderEntry
case missingTokens
case connected(expiresAtMs: Int64?)
var isConnected: Bool {
if case .connected = self { return true }
return false
}
var shortDescription: String {
switch self {
case .missingFile: "Moltbot OAuth token file not found"
case .unreadableFile: "Moltbot OAuth token file not readable"
case .invalidJSON: "Moltbot OAuth token file invalid"
case .missingProviderEntry: "No Anthropic entry in Moltbot OAuth token file"
case .missingTokens: "Anthropic entry missing tokens"
case .connected: "Moltbot OAuth credentials found"
}
}
}
static func oauthDir() -> URL {
if let override = ProcessInfo.processInfo.environment[self.clawdbotOAuthDirEnv]?
.trimmingCharacters(in: .whitespacesAndNewlines),
!override.isEmpty
{
let expanded = NSString(string: override).expandingTildeInPath
return URL(fileURLWithPath: expanded, isDirectory: true)
}
return FileManager().homeDirectoryForCurrentUser
.appendingPathComponent(".clawdbot", isDirectory: true)
.appendingPathComponent("credentials", isDirectory: true)
}
static func oauthURL() -> URL {
self.oauthDir().appendingPathComponent(self.oauthFilename)
}
static func legacyOAuthURLs() -> [URL] {
var urls: [URL] = []
let env = ProcessInfo.processInfo.environment
if let override = env[self.legacyPiDirEnv]?.trimmingCharacters(in: .whitespacesAndNewlines),
!override.isEmpty
{
let expanded = NSString(string: override).expandingTildeInPath
urls.append(URL(fileURLWithPath: expanded, isDirectory: true).appendingPathComponent(self.oauthFilename))
}
let home = FileManager().homeDirectoryForCurrentUser
urls.append(home.appendingPathComponent(".pi/agent/\(self.oauthFilename)"))
urls.append(home.appendingPathComponent(".claude/\(self.oauthFilename)"))
urls.append(home.appendingPathComponent(".config/claude/\(self.oauthFilename)"))
urls.append(home.appendingPathComponent(".config/anthropic/\(self.oauthFilename)"))
var seen = Set<String>()
return urls.filter { url in
let path = url.standardizedFileURL.path
if seen.contains(path) { return false }
seen.insert(path)
return true
}
}
static func importLegacyAnthropicOAuthIfNeeded() -> URL? {
let dest = self.oauthURL()
guard !FileManager().fileExists(atPath: dest.path) else { return nil }
for url in self.legacyOAuthURLs() {
guard FileManager().fileExists(atPath: url.path) else { continue }
guard self.anthropicOAuthStatus(at: url).isConnected else { continue }
guard let storage = self.loadStorage(at: url) else { continue }
do {
try self.saveStorage(storage)
return url
} catch {
continue
}
}
return nil
}
static func anthropicOAuthStatus() -> AnthropicOAuthStatus {
self.anthropicOAuthStatus(at: self.oauthURL())
}
static func hasAnthropicOAuth() -> Bool {
self.anthropicOAuthStatus().isConnected
}
static func anthropicOAuthStatus(at url: URL) -> AnthropicOAuthStatus {
guard FileManager().fileExists(atPath: url.path) else { return .missingFile }
guard let data = try? Data(contentsOf: url) else { return .unreadableFile }
guard let json = try? JSONSerialization.jsonObject(with: data, options: []) else { return .invalidJSON }
guard let storage = json as? [String: Any] else { return .invalidJSON }
guard let rawEntry = storage[self.providerKey] else { return .missingProviderEntry }
guard let entry = rawEntry as? [String: Any] else { return .invalidJSON }
let refresh = self.firstString(in: entry, keys: ["refresh", "refresh_token", "refreshToken"])
let access = self.firstString(in: entry, keys: ["access", "access_token", "accessToken"])
guard refresh?.isEmpty == false, access?.isEmpty == false else { return .missingTokens }
let expiresAny = entry["expires"] ?? entry["expires_at"] ?? entry["expiresAt"]
let expiresAtMs: Int64? = if let ms = expiresAny as? Int64 {
ms
} else if let number = expiresAny as? NSNumber {
number.int64Value
} else if let ms = expiresAny as? Double {
Int64(ms)
} else {
nil
}
return .connected(expiresAtMs: expiresAtMs)
}
static func loadAnthropicOAuthRefreshToken() -> String? {
let url = self.oauthURL()
guard let storage = self.loadStorage(at: url) else { return nil }
guard let rawEntry = storage[self.providerKey] as? [String: Any] else { return nil }
let refresh = self.firstString(in: rawEntry, keys: ["refresh", "refresh_token", "refreshToken"])
return refresh?.trimmingCharacters(in: .whitespacesAndNewlines)
}
private static func firstString(in dict: [String: Any], keys: [String]) -> String? {
for key in keys {
if let value = dict[key] as? String { return value }
}
return nil
}
private static func loadStorage(at url: URL) -> [String: Any]? {
guard let data = try? Data(contentsOf: url) else { return nil }
guard let json = try? JSONSerialization.jsonObject(with: data, options: []) else { return nil }
return json as? [String: Any]
}
static func saveAnthropicOAuth(_ creds: AnthropicOAuthCredentials) throws {
let url = self.oauthURL()
let existing: [String: Any] = self.loadStorage(at: url) ?? [:]
var updated = existing
updated[self.providerKey] = [
"type": creds.type,
"refresh": creds.refresh,
"access": creds.access,
"expires": creds.expires,
]
try self.saveStorage(updated)
}
private static func saveStorage(_ storage: [String: Any]) throws {
let dir = self.oauthDir()
try FileManager().createDirectory(
at: dir,
withIntermediateDirectories: true,
attributes: [.posixPermissions: 0o700])
let url = self.oauthURL()
let data = try JSONSerialization.data(
withJSONObject: storage,
options: [.prettyPrinted, .sortedKeys])
try data.write(to: url, options: [.atomic])
try FileManager().setAttributes([.posixPermissions: 0o600], ofItemAtPath: url.path)
}
}
extension Data {
fileprivate func base64URLEncodedString() -> String {
self.base64EncodedString()
.replacingOccurrences(of: "+", with: "-")
.replacingOccurrences(of: "/", with: "_")
.replacingOccurrences(of: "=", with: "")
}
}

View File

@@ -0,0 +1,216 @@
import CoreAudio
import Foundation
import OSLog
final class AudioInputDeviceObserver {
private let logger = Logger(subsystem: "bot.molt", category: "audio.devices")
private var isActive = false
private var devicesListener: AudioObjectPropertyListenerBlock?
private var defaultInputListener: AudioObjectPropertyListenerBlock?
static func defaultInputDeviceUID() -> String? {
let systemObject = AudioObjectID(kAudioObjectSystemObject)
var address = AudioObjectPropertyAddress(
mSelector: kAudioHardwarePropertyDefaultInputDevice,
mScope: kAudioObjectPropertyScopeGlobal,
mElement: kAudioObjectPropertyElementMain)
var deviceID = AudioObjectID(0)
var size = UInt32(MemoryLayout<AudioObjectID>.size)
let status = AudioObjectGetPropertyData(
systemObject,
&address,
0,
nil,
&size,
&deviceID)
guard status == noErr, deviceID != 0 else { return nil }
return self.deviceUID(for: deviceID)
}
static func aliveInputDeviceUIDs() -> Set<String> {
let systemObject = AudioObjectID(kAudioObjectSystemObject)
var address = AudioObjectPropertyAddress(
mSelector: kAudioHardwarePropertyDevices,
mScope: kAudioObjectPropertyScopeGlobal,
mElement: kAudioObjectPropertyElementMain)
var size: UInt32 = 0
var status = AudioObjectGetPropertyDataSize(systemObject, &address, 0, nil, &size)
guard status == noErr, size > 0 else { return [] }
let count = Int(size) / MemoryLayout<AudioObjectID>.size
var deviceIDs = [AudioObjectID](repeating: 0, count: count)
status = AudioObjectGetPropertyData(systemObject, &address, 0, nil, &size, &deviceIDs)
guard status == noErr else { return [] }
var output = Set<String>()
for deviceID in deviceIDs {
guard self.deviceIsAlive(deviceID) else { continue }
guard self.deviceHasInput(deviceID) else { continue }
if let uid = self.deviceUID(for: deviceID) {
output.insert(uid)
}
}
return output
}
static func defaultInputDeviceSummary() -> String {
let systemObject = AudioObjectID(kAudioObjectSystemObject)
var address = AudioObjectPropertyAddress(
mSelector: kAudioHardwarePropertyDefaultInputDevice,
mScope: kAudioObjectPropertyScopeGlobal,
mElement: kAudioObjectPropertyElementMain)
var deviceID = AudioObjectID(0)
var size = UInt32(MemoryLayout<AudioObjectID>.size)
let status = AudioObjectGetPropertyData(
systemObject,
&address,
0,
nil,
&size,
&deviceID)
guard status == noErr, deviceID != 0 else {
return "defaultInput=unknown"
}
let uid = self.deviceUID(for: deviceID) ?? "unknown"
let name = self.deviceName(for: deviceID) ?? "unknown"
return "defaultInput=\(name) (\(uid))"
}
func start(onChange: @escaping @Sendable () -> Void) {
guard !self.isActive else { return }
self.isActive = true
let systemObject = AudioObjectID(kAudioObjectSystemObject)
let queue = DispatchQueue.main
var devicesAddress = AudioObjectPropertyAddress(
mSelector: kAudioHardwarePropertyDevices,
mScope: kAudioObjectPropertyScopeGlobal,
mElement: kAudioObjectPropertyElementMain)
let devicesListener: AudioObjectPropertyListenerBlock = { _, _ in
self.logDefaultInputChange(reason: "devices")
onChange()
}
let devicesStatus = AudioObjectAddPropertyListenerBlock(
systemObject,
&devicesAddress,
queue,
devicesListener)
var defaultInputAddress = AudioObjectPropertyAddress(
mSelector: kAudioHardwarePropertyDefaultInputDevice,
mScope: kAudioObjectPropertyScopeGlobal,
mElement: kAudioObjectPropertyElementMain)
let defaultInputListener: AudioObjectPropertyListenerBlock = { _, _ in
self.logDefaultInputChange(reason: "default")
onChange()
}
let defaultStatus = AudioObjectAddPropertyListenerBlock(
systemObject,
&defaultInputAddress,
queue,
defaultInputListener)
if devicesStatus != noErr || defaultStatus != noErr {
self.logger.error("audio device observer install failed devices=\(devicesStatus) default=\(defaultStatus)")
}
self.logger.info("audio device observer started (\(Self.defaultInputDeviceSummary(), privacy: .public))")
self.devicesListener = devicesListener
self.defaultInputListener = defaultInputListener
}
func stop() {
guard self.isActive else { return }
self.isActive = false
let systemObject = AudioObjectID(kAudioObjectSystemObject)
if let devicesListener {
var devicesAddress = AudioObjectPropertyAddress(
mSelector: kAudioHardwarePropertyDevices,
mScope: kAudioObjectPropertyScopeGlobal,
mElement: kAudioObjectPropertyElementMain)
_ = AudioObjectRemovePropertyListenerBlock(
systemObject,
&devicesAddress,
DispatchQueue.main,
devicesListener)
}
if let defaultInputListener {
var defaultInputAddress = AudioObjectPropertyAddress(
mSelector: kAudioHardwarePropertyDefaultInputDevice,
mScope: kAudioObjectPropertyScopeGlobal,
mElement: kAudioObjectPropertyElementMain)
_ = AudioObjectRemovePropertyListenerBlock(
systemObject,
&defaultInputAddress,
DispatchQueue.main,
defaultInputListener)
}
self.devicesListener = nil
self.defaultInputListener = nil
}
private static func deviceUID(for deviceID: AudioObjectID) -> String? {
var address = AudioObjectPropertyAddress(
mSelector: kAudioDevicePropertyDeviceUID,
mScope: kAudioObjectPropertyScopeGlobal,
mElement: kAudioObjectPropertyElementMain)
var uid: Unmanaged<CFString>?
var size = UInt32(MemoryLayout<Unmanaged<CFString>?>.size)
let status = AudioObjectGetPropertyData(deviceID, &address, 0, nil, &size, &uid)
guard status == noErr, let uid else { return nil }
return uid.takeUnretainedValue() as String
}
private static func deviceName(for deviceID: AudioObjectID) -> String? {
var address = AudioObjectPropertyAddress(
mSelector: kAudioObjectPropertyName,
mScope: kAudioObjectPropertyScopeGlobal,
mElement: kAudioObjectPropertyElementMain)
var name: Unmanaged<CFString>?
var size = UInt32(MemoryLayout<Unmanaged<CFString>?>.size)
let status = AudioObjectGetPropertyData(deviceID, &address, 0, nil, &size, &name)
guard status == noErr, let name else { return nil }
return name.takeUnretainedValue() as String
}
private static func deviceIsAlive(_ deviceID: AudioObjectID) -> Bool {
var address = AudioObjectPropertyAddress(
mSelector: kAudioDevicePropertyDeviceIsAlive,
mScope: kAudioObjectPropertyScopeGlobal,
mElement: kAudioObjectPropertyElementMain)
var alive: UInt32 = 0
var size = UInt32(MemoryLayout<UInt32>.size)
let status = AudioObjectGetPropertyData(deviceID, &address, 0, nil, &size, &alive)
return status == noErr && alive != 0
}
private static func deviceHasInput(_ deviceID: AudioObjectID) -> Bool {
var address = AudioObjectPropertyAddress(
mSelector: kAudioDevicePropertyStreamConfiguration,
mScope: kAudioDevicePropertyScopeInput,
mElement: kAudioObjectPropertyElementMain)
var size: UInt32 = 0
var status = AudioObjectGetPropertyDataSize(deviceID, &address, 0, nil, &size)
guard status == noErr, size > 0 else { return false }
let raw = UnsafeMutableRawPointer.allocate(
byteCount: Int(size),
alignment: MemoryLayout<AudioBufferList>.alignment)
defer { raw.deallocate() }
let bufferList = raw.bindMemory(to: AudioBufferList.self, capacity: 1)
status = AudioObjectGetPropertyData(deviceID, &address, 0, nil, &size, bufferList)
guard status == noErr else { return false }
let buffers = UnsafeMutableAudioBufferListPointer(bufferList)
return buffers.contains(where: { $0.mNumberChannels > 0 })
}
private func logDefaultInputChange(reason: StaticString) {
self.logger.info("audio input changed (\(reason)) (\(Self.defaultInputDeviceSummary(), privacy: .public))")
}
}

View File

@@ -0,0 +1,84 @@
import AppKit
import Foundation
import OSLog
@MainActor
final class CLIInstallPrompter {
static let shared = CLIInstallPrompter()
private let logger = Logger(subsystem: "bot.molt", category: "cli.prompt")
private var isPrompting = false
func checkAndPromptIfNeeded(reason: String) {
guard self.shouldPrompt() else { return }
guard let version = Self.appVersion() else { return }
self.isPrompting = true
UserDefaults.standard.set(version, forKey: cliInstallPromptedVersionKey)
let alert = NSAlert()
alert.messageText = "Install Moltbot CLI?"
alert.informativeText = "Local mode needs the CLI so launchd can run the gateway."
alert.addButton(withTitle: "Install CLI")
alert.addButton(withTitle: "Not now")
alert.addButton(withTitle: "Open Settings")
let response = alert.runModal()
switch response {
case .alertFirstButtonReturn:
Task { await self.installCLI() }
case .alertThirdButtonReturn:
self.openSettings(tab: .general)
default:
break
}
self.logger.debug("cli install prompt handled reason=\(reason, privacy: .public)")
self.isPrompting = false
}
private func shouldPrompt() -> Bool {
guard !self.isPrompting else { return false }
guard AppStateStore.shared.onboardingSeen else { return false }
guard AppStateStore.shared.connectionMode == .local else { return false }
guard CLIInstaller.installedLocation() == nil else { return false }
guard let version = Self.appVersion() else { return false }
let lastPrompt = UserDefaults.standard.string(forKey: cliInstallPromptedVersionKey)
return lastPrompt != version
}
private func installCLI() async {
let status = StatusBox()
await CLIInstaller.install { message in
await status.set(message)
}
if let message = await status.get() {
let alert = NSAlert()
alert.messageText = "CLI install finished"
alert.informativeText = message
alert.runModal()
}
}
private func openSettings(tab: SettingsTab) {
SettingsTabRouter.request(tab)
SettingsWindowOpener.shared.open()
DispatchQueue.main.async {
NotificationCenter.default.post(name: .clawdbotSelectSettingsTab, object: tab)
}
}
private static func appVersion() -> String? {
Bundle.main.infoDictionary?["CFBundleShortVersionString"] as? String
}
}
private actor StatusBox {
private var value: String?
func set(_ value: String) {
self.value = value
}
func get() -> String? {
self.value
}
}

View File

@@ -0,0 +1,425 @@
import AVFoundation
import MoltbotIPC
import MoltbotKit
import CoreGraphics
import Foundation
import OSLog
actor CameraCaptureService {
struct CameraDeviceInfo: Encodable, Sendable {
let id: String
let name: String
let position: String
let deviceType: String
}
enum CameraError: LocalizedError, Sendable {
case cameraUnavailable
case microphoneUnavailable
case permissionDenied(kind: String)
case captureFailed(String)
case exportFailed(String)
var errorDescription: String? {
switch self {
case .cameraUnavailable:
"Camera unavailable"
case .microphoneUnavailable:
"Microphone unavailable"
case let .permissionDenied(kind):
"\(kind) permission denied"
case let .captureFailed(msg):
msg
case let .exportFailed(msg):
msg
}
}
}
private let logger = Logger(subsystem: "bot.molt", category: "camera")
func listDevices() -> [CameraDeviceInfo] {
Self.availableCameras().map { device in
CameraDeviceInfo(
id: device.uniqueID,
name: device.localizedName,
position: Self.positionLabel(device.position),
deviceType: device.deviceType.rawValue)
}
}
func snap(
facing: CameraFacing?,
maxWidth: Int?,
quality: Double?,
deviceId: String?,
delayMs: Int) async throws -> (data: Data, size: CGSize)
{
let facing = facing ?? .front
let normalized = Self.normalizeSnap(maxWidth: maxWidth, quality: quality)
let maxWidth = normalized.maxWidth
let quality = normalized.quality
let delayMs = max(0, delayMs)
let deviceId = deviceId?.trimmingCharacters(in: .whitespacesAndNewlines)
try await self.ensureAccess(for: .video)
let session = AVCaptureSession()
session.sessionPreset = .photo
guard let device = Self.pickCamera(facing: facing, deviceId: deviceId) else {
throw CameraError.cameraUnavailable
}
let input = try AVCaptureDeviceInput(device: device)
guard session.canAddInput(input) else {
throw CameraError.captureFailed("Failed to add camera input")
}
session.addInput(input)
let output = AVCapturePhotoOutput()
guard session.canAddOutput(output) else {
throw CameraError.captureFailed("Failed to add photo output")
}
session.addOutput(output)
output.maxPhotoQualityPrioritization = .quality
session.startRunning()
defer { session.stopRunning() }
await Self.warmUpCaptureSession()
await self.waitForExposureAndWhiteBalance(device: device)
await self.sleepDelayMs(delayMs)
let settings: AVCapturePhotoSettings = {
if output.availablePhotoCodecTypes.contains(.jpeg) {
return AVCapturePhotoSettings(format: [AVVideoCodecKey: AVVideoCodecType.jpeg])
}
return AVCapturePhotoSettings()
}()
settings.photoQualityPrioritization = .quality
var delegate: PhotoCaptureDelegate?
let rawData: Data = try await withCheckedThrowingContinuation { cont in
let d = PhotoCaptureDelegate(cont)
delegate = d
output.capturePhoto(with: settings, delegate: d)
}
withExtendedLifetime(delegate) {}
let maxPayloadBytes = 5 * 1024 * 1024
// Base64 inflates payloads by ~4/3; cap encoded bytes so the payload stays under 5MB (API limit).
let maxEncodedBytes = (maxPayloadBytes / 4) * 3
let res = try JPEGTranscoder.transcodeToJPEG(
imageData: rawData,
maxWidthPx: maxWidth,
quality: quality,
maxBytes: maxEncodedBytes)
return (data: res.data, size: CGSize(width: res.widthPx, height: res.heightPx))
}
func clip(
facing: CameraFacing?,
durationMs: Int?,
includeAudio: Bool,
deviceId: String?,
outPath: String?) async throws -> (path: String, durationMs: Int, hasAudio: Bool)
{
let facing = facing ?? .front
let durationMs = Self.clampDurationMs(durationMs)
let deviceId = deviceId?.trimmingCharacters(in: .whitespacesAndNewlines)
try await self.ensureAccess(for: .video)
if includeAudio {
try await self.ensureAccess(for: .audio)
}
let session = AVCaptureSession()
session.sessionPreset = .high
guard let camera = Self.pickCamera(facing: facing, deviceId: deviceId) else {
throw CameraError.cameraUnavailable
}
let cameraInput = try AVCaptureDeviceInput(device: camera)
guard session.canAddInput(cameraInput) else {
throw CameraError.captureFailed("Failed to add camera input")
}
session.addInput(cameraInput)
if includeAudio {
guard let mic = AVCaptureDevice.default(for: .audio) else {
throw CameraError.microphoneUnavailable
}
let micInput = try AVCaptureDeviceInput(device: mic)
guard session.canAddInput(micInput) else {
throw CameraError.captureFailed("Failed to add microphone input")
}
session.addInput(micInput)
}
let output = AVCaptureMovieFileOutput()
guard session.canAddOutput(output) else {
throw CameraError.captureFailed("Failed to add movie output")
}
session.addOutput(output)
output.maxRecordedDuration = CMTime(value: Int64(durationMs), timescale: 1000)
session.startRunning()
defer { session.stopRunning() }
await Self.warmUpCaptureSession()
let tmpMovURL = FileManager().temporaryDirectory
.appendingPathComponent("moltbot-camera-\(UUID().uuidString).mov")
defer { try? FileManager().removeItem(at: tmpMovURL) }
let outputURL: URL = {
if let outPath, !outPath.trimmingCharacters(in: .whitespacesAndNewlines).isEmpty {
return URL(fileURLWithPath: outPath)
}
return FileManager().temporaryDirectory
.appendingPathComponent("moltbot-camera-\(UUID().uuidString).mp4")
}()
// Ensure we don't fail exporting due to an existing file.
try? FileManager().removeItem(at: outputURL)
let logger = self.logger
var delegate: MovieFileDelegate?
let recordedURL: URL = try await withCheckedThrowingContinuation { cont in
let d = MovieFileDelegate(cont, logger: logger)
delegate = d
output.startRecording(to: tmpMovURL, recordingDelegate: d)
}
withExtendedLifetime(delegate) {}
try await Self.exportToMP4(inputURL: recordedURL, outputURL: outputURL)
return (path: outputURL.path, durationMs: durationMs, hasAudio: includeAudio)
}
private func ensureAccess(for mediaType: AVMediaType) async throws {
let status = AVCaptureDevice.authorizationStatus(for: mediaType)
switch status {
case .authorized:
return
case .notDetermined:
let ok = await withCheckedContinuation(isolation: nil) { cont in
AVCaptureDevice.requestAccess(for: mediaType) { granted in
cont.resume(returning: granted)
}
}
if !ok {
throw CameraError.permissionDenied(kind: mediaType == .video ? "Camera" : "Microphone")
}
case .denied, .restricted:
throw CameraError.permissionDenied(kind: mediaType == .video ? "Camera" : "Microphone")
@unknown default:
throw CameraError.permissionDenied(kind: mediaType == .video ? "Camera" : "Microphone")
}
}
private nonisolated static func availableCameras() -> [AVCaptureDevice] {
var types: [AVCaptureDevice.DeviceType] = [
.builtInWideAngleCamera,
.continuityCamera,
]
if let external = externalDeviceType() {
types.append(external)
}
let session = AVCaptureDevice.DiscoverySession(
deviceTypes: types,
mediaType: .video,
position: .unspecified)
return session.devices
}
private nonisolated static func externalDeviceType() -> AVCaptureDevice.DeviceType? {
if #available(macOS 14.0, *) {
return .external
}
// Use raw value to avoid deprecated symbol in the SDK.
return AVCaptureDevice.DeviceType(rawValue: "AVCaptureDeviceTypeExternalUnknown")
}
private nonisolated static func pickCamera(
facing: CameraFacing,
deviceId: String?) -> AVCaptureDevice?
{
if let deviceId, !deviceId.isEmpty {
if let match = availableCameras().first(where: { $0.uniqueID == deviceId }) {
return match
}
}
let position: AVCaptureDevice.Position = (facing == .front) ? .front : .back
if let device = AVCaptureDevice.default(.builtInWideAngleCamera, for: .video, position: position) {
return device
}
// Many macOS cameras report `unspecified` position; fall back to any default.
return AVCaptureDevice.default(for: .video)
}
private nonisolated static func clampQuality(_ quality: Double?) -> Double {
let q = quality ?? 0.9
return min(1.0, max(0.05, q))
}
nonisolated static func normalizeSnap(maxWidth: Int?, quality: Double?) -> (maxWidth: Int, quality: Double) {
// Default to a reasonable max width to keep downstream payload sizes manageable.
// If you need full-res, explicitly request a larger maxWidth.
let maxWidth = maxWidth.flatMap { $0 > 0 ? $0 : nil } ?? 1600
let quality = Self.clampQuality(quality)
return (maxWidth: maxWidth, quality: quality)
}
private nonisolated static func clampDurationMs(_ ms: Int?) -> Int {
let v = ms ?? 3000
return min(60000, max(250, v))
}
private nonisolated static func exportToMP4(inputURL: URL, outputURL: URL) async throws {
let asset = AVURLAsset(url: inputURL)
guard let export = AVAssetExportSession(asset: asset, presetName: AVAssetExportPresetMediumQuality) else {
throw CameraError.exportFailed("Failed to create export session")
}
export.shouldOptimizeForNetworkUse = true
if #available(macOS 15.0, *) {
do {
try await export.export(to: outputURL, as: .mp4)
return
} catch {
throw CameraError.exportFailed(error.localizedDescription)
}
} else {
export.outputURL = outputURL
export.outputFileType = .mp4
try await withCheckedThrowingContinuation(isolation: nil) { (cont: CheckedContinuation<Void, Error>) in
export.exportAsynchronously {
cont.resume(returning: ())
}
}
switch export.status {
case .completed:
return
case .failed:
throw CameraError.exportFailed(export.error?.localizedDescription ?? "export failed")
case .cancelled:
throw CameraError.exportFailed("export cancelled")
default:
throw CameraError.exportFailed("export did not complete (\(export.status.rawValue))")
}
}
}
private nonisolated static func warmUpCaptureSession() async {
// A short delay after `startRunning()` significantly reduces "blank first frame" captures on some devices.
try? await Task.sleep(nanoseconds: 150_000_000) // 150ms
}
private func waitForExposureAndWhiteBalance(device: AVCaptureDevice) async {
let stepNs: UInt64 = 50_000_000
let maxSteps = 30 // ~1.5s
for _ in 0..<maxSteps {
if !(device.isAdjustingExposure || device.isAdjustingWhiteBalance) {
return
}
try? await Task.sleep(nanoseconds: stepNs)
}
}
private func sleepDelayMs(_ delayMs: Int) async {
guard delayMs > 0 else { return }
let ns = UInt64(min(delayMs, 10000)) * 1_000_000
try? await Task.sleep(nanoseconds: ns)
}
private nonisolated static func positionLabel(_ position: AVCaptureDevice.Position) -> String {
switch position {
case .front: "front"
case .back: "back"
default: "unspecified"
}
}
}
private final class PhotoCaptureDelegate: NSObject, AVCapturePhotoCaptureDelegate {
private var cont: CheckedContinuation<Data, Error>?
private var didResume = false
init(_ cont: CheckedContinuation<Data, Error>) {
self.cont = cont
}
func photoOutput(
_ output: AVCapturePhotoOutput,
didFinishProcessingPhoto photo: AVCapturePhoto,
error: Error?)
{
guard !self.didResume, let cont else { return }
self.didResume = true
self.cont = nil
if let error {
cont.resume(throwing: error)
return
}
guard let data = photo.fileDataRepresentation() else {
cont.resume(throwing: CameraCaptureService.CameraError.captureFailed("No photo data"))
return
}
if data.isEmpty {
cont.resume(throwing: CameraCaptureService.CameraError.captureFailed("Photo data empty"))
return
}
cont.resume(returning: data)
}
func photoOutput(
_ output: AVCapturePhotoOutput,
didFinishCaptureFor resolvedSettings: AVCaptureResolvedPhotoSettings,
error: Error?)
{
guard let error else { return }
guard !self.didResume, let cont else { return }
self.didResume = true
self.cont = nil
cont.resume(throwing: error)
}
}
private final class MovieFileDelegate: NSObject, AVCaptureFileOutputRecordingDelegate {
private var cont: CheckedContinuation<URL, Error>?
private let logger: Logger
init(_ cont: CheckedContinuation<URL, Error>, logger: Logger) {
self.cont = cont
self.logger = logger
}
func fileOutput(
_ output: AVCaptureFileOutput,
didFinishRecordingTo outputFileURL: URL,
from connections: [AVCaptureConnection],
error: Error?)
{
guard let cont else { return }
self.cont = nil
if let error {
let ns = error as NSError
if ns.domain == AVFoundationErrorDomain,
ns.code == AVError.maximumDurationReached.rawValue
{
cont.resume(returning: outputFileURL)
return
}
self.logger.error("camera record failed: \(error.localizedDescription, privacy: .public)")
cont.resume(throwing: error)
return
}
cont.resume(returning: outputFileURL)
}
}

View File

@@ -0,0 +1,94 @@
import CoreServices
import Foundation
final class CanvasFileWatcher: @unchecked Sendable {
private let url: URL
private let queue: DispatchQueue
private var stream: FSEventStreamRef?
private var pending = false
private let onChange: () -> Void
init(url: URL, onChange: @escaping () -> Void) {
self.url = url
self.queue = DispatchQueue(label: "bot.molt.canvaswatcher")
self.onChange = onChange
}
deinit {
self.stop()
}
func start() {
guard self.stream == nil else { return }
let retainedSelf = Unmanaged.passRetained(self)
var context = FSEventStreamContext(
version: 0,
info: retainedSelf.toOpaque(),
retain: nil,
release: { pointer in
guard let pointer else { return }
Unmanaged<CanvasFileWatcher>.fromOpaque(pointer).release()
},
copyDescription: nil)
let paths = [self.url.path] as CFArray
let flags = FSEventStreamCreateFlags(
kFSEventStreamCreateFlagFileEvents |
kFSEventStreamCreateFlagUseCFTypes |
kFSEventStreamCreateFlagNoDefer)
guard let stream = FSEventStreamCreate(
kCFAllocatorDefault,
Self.callback,
&context,
paths,
FSEventStreamEventId(kFSEventStreamEventIdSinceNow),
0.05,
flags)
else {
retainedSelf.release()
return
}
self.stream = stream
FSEventStreamSetDispatchQueue(stream, self.queue)
if FSEventStreamStart(stream) == false {
self.stream = nil
FSEventStreamSetDispatchQueue(stream, nil)
FSEventStreamInvalidate(stream)
FSEventStreamRelease(stream)
}
}
func stop() {
guard let stream = self.stream else { return }
self.stream = nil
FSEventStreamStop(stream)
FSEventStreamSetDispatchQueue(stream, nil)
FSEventStreamInvalidate(stream)
FSEventStreamRelease(stream)
}
}
extension CanvasFileWatcher {
private static let callback: FSEventStreamCallback = { _, info, numEvents, _, eventFlags, _ in
guard let info else { return }
let watcher = Unmanaged<CanvasFileWatcher>.fromOpaque(info).takeUnretainedValue()
watcher.handleEvents(numEvents: numEvents, eventFlags: eventFlags)
}
private func handleEvents(numEvents: Int, eventFlags: UnsafePointer<FSEventStreamEventFlags>?) {
guard numEvents > 0 else { return }
guard eventFlags != nil else { return }
// Coalesce rapid changes (common during builds/atomic saves).
if self.pending { return }
self.pending = true
self.queue.asyncAfter(deadline: .now() + 0.12) { [weak self] in
guard let self else { return }
self.pending = false
self.onChange()
}
}
}

View File

@@ -0,0 +1,342 @@
import AppKit
import MoltbotIPC
import MoltbotKit
import Foundation
import OSLog
@MainActor
final class CanvasManager {
static let shared = CanvasManager()
private static let logger = Logger(subsystem: "bot.molt", category: "CanvasManager")
private var panelController: CanvasWindowController?
private var panelSessionKey: String?
private var lastAutoA2UIUrl: String?
private var gatewayWatchTask: Task<Void, Never>?
private init() {
self.startGatewayObserver()
}
var onPanelVisibilityChanged: ((Bool) -> Void)?
/// Optional anchor provider (e.g. menu bar status item). If nil, Canvas anchors to the mouse cursor.
var defaultAnchorProvider: (() -> NSRect?)?
private nonisolated static let canvasRoot: URL = {
let base = FileManager().urls(for: .applicationSupportDirectory, in: .userDomainMask).first!
return base.appendingPathComponent("Moltbot/canvas", isDirectory: true)
}()
func show(sessionKey: String, path: String? = nil, placement: CanvasPlacement? = nil) throws -> String {
try self.showDetailed(sessionKey: sessionKey, target: path, placement: placement).directory
}
func showDetailed(
sessionKey: String,
target: String? = nil,
placement: CanvasPlacement? = nil) throws -> CanvasShowResult
{
Self.logger.debug(
"""
showDetailed start session=\(sessionKey, privacy: .public) \
target=\(target ?? "", privacy: .public) \
placement=\(placement != nil)
""")
let anchorProvider = self.defaultAnchorProvider ?? Self.mouseAnchorProvider
let session = sessionKey.trimmingCharacters(in: .whitespacesAndNewlines)
let normalizedTarget = target?
.trimmingCharacters(in: .whitespacesAndNewlines)
.nonEmpty
if let controller = self.panelController, self.panelSessionKey == session {
Self.logger.debug("showDetailed reuse existing session=\(session, privacy: .public)")
controller.onVisibilityChanged = { [weak self] visible in
self?.onPanelVisibilityChanged?(visible)
}
controller.presentAnchoredPanel(anchorProvider: anchorProvider)
controller.applyPreferredPlacement(placement)
self.refreshDebugStatus()
// Existing session: only navigate when an explicit target was provided.
if let normalizedTarget {
controller.load(target: normalizedTarget)
return self.makeShowResult(
directory: controller.directoryPath,
target: target,
effectiveTarget: normalizedTarget)
}
self.maybeAutoNavigateToA2UIAsync(controller: controller)
return CanvasShowResult(
directory: controller.directoryPath,
target: target,
effectiveTarget: nil,
status: .shown,
url: nil)
}
Self.logger.debug("showDetailed creating new session=\(session, privacy: .public)")
self.panelController?.close()
self.panelController = nil
self.panelSessionKey = nil
Self.logger.debug("showDetailed ensure canvas root dir")
try FileManager().createDirectory(at: Self.canvasRoot, withIntermediateDirectories: true)
Self.logger.debug("showDetailed init CanvasWindowController")
let controller = try CanvasWindowController(
sessionKey: session,
root: Self.canvasRoot,
presentation: .panel(anchorProvider: anchorProvider))
Self.logger.debug("showDetailed CanvasWindowController init done")
controller.onVisibilityChanged = { [weak self] visible in
self?.onPanelVisibilityChanged?(visible)
}
self.panelController = controller
self.panelSessionKey = session
controller.applyPreferredPlacement(placement)
// New session: default to "/" so the user sees either the welcome page or `index.html`.
let effectiveTarget = normalizedTarget ?? "/"
Self.logger.debug("showDetailed showCanvas effectiveTarget=\(effectiveTarget, privacy: .public)")
controller.showCanvas(path: effectiveTarget)
Self.logger.debug("showDetailed showCanvas done")
if normalizedTarget == nil {
self.maybeAutoNavigateToA2UIAsync(controller: controller)
}
self.refreshDebugStatus()
return self.makeShowResult(
directory: controller.directoryPath,
target: target,
effectiveTarget: effectiveTarget)
}
func hide(sessionKey: String) {
let session = sessionKey.trimmingCharacters(in: .whitespacesAndNewlines)
guard self.panelSessionKey == session else { return }
self.panelController?.hideCanvas()
}
func hideAll() {
self.panelController?.hideCanvas()
}
func eval(sessionKey: String, javaScript: String) async throws -> String {
_ = try self.show(sessionKey: sessionKey, path: nil)
guard let controller = self.panelController else { return "" }
return try await controller.eval(javaScript: javaScript)
}
func snapshot(sessionKey: String, outPath: String?) async throws -> String {
_ = try self.show(sessionKey: sessionKey, path: nil)
guard let controller = self.panelController else {
throw NSError(domain: "Canvas", code: 21, userInfo: [NSLocalizedDescriptionKey: "canvas not available"])
}
return try await controller.snapshot(to: outPath)
}
// MARK: - Gateway A2UI auto-nav
private func startGatewayObserver() {
self.gatewayWatchTask?.cancel()
self.gatewayWatchTask = Task { [weak self] in
guard let self else { return }
let stream = await GatewayConnection.shared.subscribe(bufferingNewest: 1)
for await push in stream {
self.handleGatewayPush(push)
}
}
}
private func handleGatewayPush(_ push: GatewayPush) {
guard case let .snapshot(snapshot) = push else { return }
let raw = snapshot.canvashosturl?.trimmingCharacters(in: .whitespacesAndNewlines) ?? ""
if raw.isEmpty {
Self.logger.debug("canvas host url missing in gateway snapshot")
} else {
Self.logger.debug("canvas host url snapshot=\(raw, privacy: .public)")
}
let a2uiUrl = Self.resolveA2UIHostUrl(from: raw)
if a2uiUrl == nil, !raw.isEmpty {
Self.logger.debug("canvas host url invalid; cannot resolve A2UI")
}
guard let controller = self.panelController else {
if a2uiUrl != nil {
Self.logger.debug("canvas panel not visible; skipping auto-nav")
}
return
}
self.maybeAutoNavigateToA2UI(controller: controller, a2uiUrl: a2uiUrl)
}
private func maybeAutoNavigateToA2UIAsync(controller: CanvasWindowController) {
Task { [weak self] in
guard let self else { return }
let a2uiUrl = await self.resolveA2UIHostUrl()
await MainActor.run {
guard self.panelController === controller else { return }
self.maybeAutoNavigateToA2UI(controller: controller, a2uiUrl: a2uiUrl)
}
}
}
private func maybeAutoNavigateToA2UI(controller: CanvasWindowController, a2uiUrl: String?) {
guard let a2uiUrl else { return }
let shouldNavigate = controller.shouldAutoNavigateToA2UI(lastAutoTarget: self.lastAutoA2UIUrl)
guard shouldNavigate else {
Self.logger.debug("canvas auto-nav skipped; target unchanged")
return
}
Self.logger.debug("canvas auto-nav -> \(a2uiUrl, privacy: .public)")
controller.load(target: a2uiUrl)
self.lastAutoA2UIUrl = a2uiUrl
}
private func resolveA2UIHostUrl() async -> String? {
let raw = await GatewayConnection.shared.canvasHostUrl()
return Self.resolveA2UIHostUrl(from: raw)
}
func refreshDebugStatus() {
guard let controller = self.panelController else { return }
let enabled = AppStateStore.shared.debugPaneEnabled
let mode = AppStateStore.shared.connectionMode
let title: String?
let subtitle: String?
switch mode {
case .remote:
title = "Remote control"
switch ControlChannel.shared.state {
case .connected:
subtitle = "Connected"
case .connecting:
subtitle = "Connecting…"
case .disconnected:
subtitle = "Disconnected"
case let .degraded(message):
subtitle = message.isEmpty ? "Degraded" : message
}
case .local:
title = GatewayProcessManager.shared.status.label
subtitle = mode.rawValue
case .unconfigured:
title = "Unconfigured"
subtitle = mode.rawValue
}
controller.updateDebugStatus(enabled: enabled, title: title, subtitle: subtitle)
}
private static func resolveA2UIHostUrl(from raw: String?) -> String? {
let trimmed = raw?.trimmingCharacters(in: .whitespacesAndNewlines) ?? ""
guard !trimmed.isEmpty, let base = URL(string: trimmed) else { return nil }
return base.appendingPathComponent("__moltbot__/a2ui/").absoluteString + "?platform=macos"
}
// MARK: - Anchoring
private static func mouseAnchorProvider() -> NSRect? {
let pt = NSEvent.mouseLocation
return NSRect(x: pt.x, y: pt.y, width: 1, height: 1)
}
// placement interpretation is handled by the window controller.
// MARK: - Helpers
private static func directURL(for target: String?) -> URL? {
guard let target else { return nil }
let trimmed = target.trimmingCharacters(in: .whitespacesAndNewlines)
guard !trimmed.isEmpty else { return nil }
if let url = URL(string: trimmed), let scheme = url.scheme?.lowercased() {
if scheme == "https" || scheme == "http" || scheme == "file" { return url }
}
// Convenience: existing absolute *file* paths resolve as local files.
// (Avoid treating Canvas routes like "/" as filesystem paths.)
if trimmed.hasPrefix("/") {
var isDir: ObjCBool = false
if FileManager().fileExists(atPath: trimmed, isDirectory: &isDir), !isDir.boolValue {
return URL(fileURLWithPath: trimmed)
}
}
return nil
}
private func makeShowResult(
directory: String,
target: String?,
effectiveTarget: String) -> CanvasShowResult
{
if let url = Self.directURL(for: effectiveTarget) {
return CanvasShowResult(
directory: directory,
target: target,
effectiveTarget: effectiveTarget,
status: .web,
url: url.absoluteString)
}
let sessionDir = URL(fileURLWithPath: directory)
let status = Self.localStatus(sessionDir: sessionDir, target: effectiveTarget)
let host = sessionDir.lastPathComponent
let canvasURL = CanvasScheme.makeURL(session: host, path: effectiveTarget)?.absoluteString
return CanvasShowResult(
directory: directory,
target: target,
effectiveTarget: effectiveTarget,
status: status,
url: canvasURL)
}
private static func localStatus(sessionDir: URL, target: String) -> CanvasShowStatus {
let fm = FileManager()
let trimmed = target.trimmingCharacters(in: .whitespacesAndNewlines)
let withoutQuery = trimmed.split(separator: "?", maxSplits: 1, omittingEmptySubsequences: false).first
.map(String.init) ?? trimmed
var path = withoutQuery
if path.hasPrefix("/") { path.removeFirst() }
path = path.removingPercentEncoding ?? path
// Root special-case: built-in scaffold page when no index exists.
if path.isEmpty {
let a = sessionDir.appendingPathComponent("index.html", isDirectory: false)
let b = sessionDir.appendingPathComponent("index.htm", isDirectory: false)
if fm.fileExists(atPath: a.path) || fm.fileExists(atPath: b.path) { return .ok }
return .welcome
}
// Direct file or directory.
var candidate = sessionDir.appendingPathComponent(path, isDirectory: false)
var isDir: ObjCBool = false
if fm.fileExists(atPath: candidate.path, isDirectory: &isDir) {
if isDir.boolValue {
return Self.indexExists(in: candidate) ? .ok : .notFound
}
return .ok
}
// Directory index behavior ("/yolo" -> "yolo/index.html") if directory exists.
if !path.isEmpty, !path.hasSuffix("/") {
candidate = sessionDir.appendingPathComponent(path, isDirectory: true)
if fm.fileExists(atPath: candidate.path, isDirectory: &isDir), isDir.boolValue {
return Self.indexExists(in: candidate) ? .ok : .notFound
}
}
return .notFound
}
private static func indexExists(in dir: URL) -> Bool {
let fm = FileManager()
let a = dir.appendingPathComponent("index.html", isDirectory: false)
if fm.fileExists(atPath: a.path) { return true }
let b = dir.appendingPathComponent("index.htm", isDirectory: false)
return fm.fileExists(atPath: b.path)
}
// no bundled A2UI shell; scaffold fallback is purely visual
}

View File

@@ -0,0 +1,259 @@
import MoltbotKit
import Foundation
import OSLog
import WebKit
private let canvasLogger = Logger(subsystem: "bot.molt", category: "Canvas")
final class CanvasSchemeHandler: NSObject, WKURLSchemeHandler {
private let root: URL
init(root: URL) {
self.root = root
}
func webView(_: WKWebView, start urlSchemeTask: WKURLSchemeTask) {
guard let url = urlSchemeTask.request.url else {
urlSchemeTask.didFailWithError(NSError(domain: "Canvas", code: 1, userInfo: [
NSLocalizedDescriptionKey: "missing url",
]))
return
}
let response = self.response(for: url)
let mime = response.mime
let data = response.data
let encoding = self.textEncodingName(forMimeType: mime)
let urlResponse = URLResponse(
url: url,
mimeType: mime,
expectedContentLength: data.count,
textEncodingName: encoding)
urlSchemeTask.didReceive(urlResponse)
urlSchemeTask.didReceive(data)
urlSchemeTask.didFinish()
}
func webView(_: WKWebView, stop _: WKURLSchemeTask) {
// no-op
}
private struct CanvasResponse {
let mime: String
let data: Data
}
private func response(for url: URL) -> CanvasResponse {
guard url.scheme == CanvasScheme.scheme else {
return self.html("Invalid scheme.")
}
guard let session = url.host, !session.isEmpty else {
return self.html("Missing session.")
}
// Keep session component safe; don't allow slashes or traversal.
if session.contains("/") || session.contains("..") {
return self.html("Invalid session.")
}
let sessionRoot = self.root.appendingPathComponent(session, isDirectory: true)
// Path mapping: request path maps directly into the session dir.
var path = url.path
if let qIdx = path.firstIndex(of: "?") { path = String(path[..<qIdx]) }
if path.hasPrefix("/") { path.removeFirst() }
path = path.removingPercentEncoding ?? path
// Special-case: welcome page when root index is missing.
if path.isEmpty {
let indexA = sessionRoot.appendingPathComponent("index.html", isDirectory: false)
let indexB = sessionRoot.appendingPathComponent("index.htm", isDirectory: false)
if !FileManager().fileExists(atPath: indexA.path),
!FileManager().fileExists(atPath: indexB.path)
{
return self.scaffoldPage(sessionRoot: sessionRoot)
}
}
let resolved = self.resolveFileURL(sessionRoot: sessionRoot, requestPath: path)
guard let fileURL = resolved else {
return self.html("Not Found", title: "Canvas: 404")
}
// Directory traversal guard: served files must live under the session root.
let standardizedRoot = sessionRoot.standardizedFileURL
let standardizedFile = fileURL.standardizedFileURL
guard standardizedFile.path.hasPrefix(standardizedRoot.path) else {
return self.html("Forbidden", title: "Canvas: 403")
}
do {
let data = try Data(contentsOf: standardizedFile)
let mime = CanvasScheme.mimeType(forExtension: standardizedFile.pathExtension)
let servedPath = standardizedFile.path
canvasLogger.debug(
"served \(session, privacy: .public)/\(path, privacy: .public) -> \(servedPath, privacy: .public)")
return CanvasResponse(mime: mime, data: data)
} catch {
let failedPath = standardizedFile.path
let errorText = error.localizedDescription
canvasLogger
.error(
"failed reading \(failedPath, privacy: .public): \(errorText, privacy: .public)")
return self.html("Failed to read file.", title: "Canvas error")
}
}
private func resolveFileURL(sessionRoot: URL, requestPath: String) -> URL? {
let fm = FileManager()
var candidate = sessionRoot.appendingPathComponent(requestPath, isDirectory: false)
var isDir: ObjCBool = false
if fm.fileExists(atPath: candidate.path, isDirectory: &isDir) {
if isDir.boolValue {
if let idx = self.resolveIndex(in: candidate) { return idx }
return nil
}
return candidate
}
// Directory index behavior:
// - "/yolo" serves "<yolo>/index.html" if that directory exists.
if !requestPath.isEmpty, !requestPath.hasSuffix("/") {
candidate = sessionRoot.appendingPathComponent(requestPath, isDirectory: true)
if fm.fileExists(atPath: candidate.path, isDirectory: &isDir), isDir.boolValue {
if let idx = self.resolveIndex(in: candidate) { return idx }
}
}
// Root fallback:
// - "/" serves "<sessionRoot>/index.html" if present.
if requestPath.isEmpty {
return self.resolveIndex(in: sessionRoot)
}
return nil
}
private func resolveIndex(in dir: URL) -> URL? {
let fm = FileManager()
let a = dir.appendingPathComponent("index.html", isDirectory: false)
if fm.fileExists(atPath: a.path) { return a }
let b = dir.appendingPathComponent("index.htm", isDirectory: false)
if fm.fileExists(atPath: b.path) { return b }
return nil
}
private func html(_ body: String, title: String = "Canvas") -> CanvasResponse {
let html = """
<!doctype html>
<html>
<head>
<meta charset="utf-8" />
<meta name="viewport" content="width=device-width, initial-scale=1" />
<title>\(title)</title>
<style>
:root { color-scheme: light; }
html,body { height:100%; margin:0; }
body {
font: 13px -apple-system, system-ui;
display:flex;
align-items:center;
justify-content:center;
background: #fff;
color:#111827;
}
.card {
max-width: 520px;
padding: 18px 18px;
border-radius: 12px;
border: 1px solid rgba(0,0,0,.08);
box-shadow: 0 10px 30px rgba(0,0,0,.08);
}
.muted { color:#6b7280; margin-top:8px; }
code { font-family: ui-monospace, SFMono-Regular, Menlo, monospace; }
</style>
</head>
<body>
<div class="card">
<div>\(body)</div>
</div>
</body>
</html>
"""
return CanvasResponse(mime: "text/html", data: Data(html.utf8))
}
private func welcomePage(sessionRoot: URL) -> CanvasResponse {
let escaped = sessionRoot.path
.replacingOccurrences(of: "&", with: "&amp;")
.replacingOccurrences(of: "<", with: "&lt;")
.replacingOccurrences(of: ">", with: "&gt;")
let body = """
<div style="font-weight:600; font-size:14px;">Canvas is ready.</div>
<div class="muted">Create <code>index.html</code> in:</div>
<div style="margin-top:10px;"><code>\(escaped)</code></div>
"""
return self.html(body, title: "Canvas")
}
private func scaffoldPage(sessionRoot: URL) -> CanvasResponse {
// Default Canvas UX: when no index exists, show the built-in scaffold page.
if let data = self.loadBundledResourceData(relativePath: "CanvasScaffold/scaffold.html") {
return CanvasResponse(mime: "text/html", data: data)
}
// Fallback for dev misconfiguration: show the classic welcome page.
return self.welcomePage(sessionRoot: sessionRoot)
}
private func loadBundledResourceData(relativePath: String) -> Data? {
let trimmed = relativePath.trimmingCharacters(in: .whitespacesAndNewlines)
guard !trimmed.isEmpty else { return nil }
if trimmed.contains("..") || trimmed.contains("\\") { return nil }
let parts = trimmed.split(separator: "/")
guard let filename = parts.last else { return nil }
let subdirectory =
parts.count > 1 ? parts.dropLast().joined(separator: "/") : nil
let fileURL = URL(fileURLWithPath: String(filename))
let ext = fileURL.pathExtension
let name = fileURL.deletingPathExtension().lastPathComponent
guard !name.isEmpty, !ext.isEmpty else { return nil }
let bundle = MoltbotKitResources.bundle
let resourceURL =
bundle.url(forResource: name, withExtension: ext, subdirectory: subdirectory)
?? bundle.url(forResource: name, withExtension: ext)
guard let resourceURL else { return nil }
return try? Data(contentsOf: resourceURL)
}
private func textEncodingName(forMimeType mimeType: String) -> String? {
if mimeType.hasPrefix("text/") { return "utf-8" }
switch mimeType {
case "application/javascript", "application/json", "image/svg+xml":
return "utf-8"
default:
return nil
}
}
}
#if DEBUG
extension CanvasSchemeHandler {
func _testResponse(for url: URL) -> (mime: String, data: Data) {
let response = self.response(for: url)
return (response.mime, response.data)
}
func _testResolveFileURL(sessionRoot: URL, requestPath: String) -> URL? {
self.resolveFileURL(sessionRoot: sessionRoot, requestPath: requestPath)
}
func _testTextEncodingName(for mimeType: String) -> String? {
self.textEncodingName(forMimeType: mimeType)
}
}
#endif

View File

@@ -0,0 +1,26 @@
import AppKit
let canvasWindowLogger = Logger(subsystem: "bot.molt", category: "Canvas")
enum CanvasLayout {
static let panelSize = NSSize(width: 520, height: 680)
static let windowSize = NSSize(width: 1120, height: 840)
static let anchorPadding: CGFloat = 8
static let defaultPadding: CGFloat = 10
static let minPanelSize = NSSize(width: 360, height: 360)
}
final class CanvasPanel: NSPanel {
override var canBecomeKey: Bool { true }
override var canBecomeMain: Bool { true }
}
enum CanvasPresentation {
case window
case panel(anchorProvider: () -> NSRect?)
var isPanel: Bool {
if case .panel = self { return true }
return false
}
}

View File

@@ -0,0 +1,217 @@
import MoltbotProtocol
import Foundation
enum MoltbotConfigFile {
private static let logger = Logger(subsystem: "bot.molt", category: "config")
static func url() -> URL {
MoltbotPaths.configURL
}
static func stateDirURL() -> URL {
MoltbotPaths.stateDirURL
}
static func defaultWorkspaceURL() -> URL {
MoltbotPaths.workspaceURL
}
static func loadDict() -> [String: Any] {
let url = self.url()
guard FileManager().fileExists(atPath: url.path) else { return [:] }
do {
let data = try Data(contentsOf: url)
guard let root = self.parseConfigData(data) else {
self.logger.warning("config JSON root invalid")
return [:]
}
return root
} catch {
self.logger.warning("config read failed: \(error.localizedDescription)")
return [:]
}
}
static func saveDict(_ dict: [String: Any]) {
// Nix mode disables config writes in production, but tests rely on saving temp configs.
if ProcessInfo.processInfo.isNixMode, !ProcessInfo.processInfo.isRunningTests { return }
do {
let data = try JSONSerialization.data(withJSONObject: dict, options: [.prettyPrinted, .sortedKeys])
let url = self.url()
try FileManager().createDirectory(
at: url.deletingLastPathComponent(),
withIntermediateDirectories: true)
try data.write(to: url, options: [.atomic])
} catch {
self.logger.error("config save failed: \(error.localizedDescription)")
}
}
static func loadGatewayDict() -> [String: Any] {
let root = self.loadDict()
return root["gateway"] as? [String: Any] ?? [:]
}
static func updateGatewayDict(_ mutate: (inout [String: Any]) -> Void) {
var root = self.loadDict()
var gateway = root["gateway"] as? [String: Any] ?? [:]
mutate(&gateway)
if gateway.isEmpty {
root.removeValue(forKey: "gateway")
} else {
root["gateway"] = gateway
}
self.saveDict(root)
}
static func browserControlEnabled(defaultValue: Bool = true) -> Bool {
let root = self.loadDict()
let browser = root["browser"] as? [String: Any]
return browser?["enabled"] as? Bool ?? defaultValue
}
static func setBrowserControlEnabled(_ enabled: Bool) {
var root = self.loadDict()
var browser = root["browser"] as? [String: Any] ?? [:]
browser["enabled"] = enabled
root["browser"] = browser
self.saveDict(root)
self.logger.debug("browser control updated enabled=\(enabled)")
}
static func agentWorkspace() -> String? {
let root = self.loadDict()
let agents = root["agents"] as? [String: Any]
let defaults = agents?["defaults"] as? [String: Any]
return defaults?["workspace"] as? String
}
static func setAgentWorkspace(_ workspace: String?) {
var root = self.loadDict()
var agents = root["agents"] as? [String: Any] ?? [:]
var defaults = agents["defaults"] as? [String: Any] ?? [:]
let trimmed = workspace?.trimmingCharacters(in: .whitespacesAndNewlines) ?? ""
if trimmed.isEmpty {
defaults.removeValue(forKey: "workspace")
} else {
defaults["workspace"] = trimmed
}
if defaults.isEmpty {
agents.removeValue(forKey: "defaults")
} else {
agents["defaults"] = defaults
}
if agents.isEmpty {
root.removeValue(forKey: "agents")
} else {
root["agents"] = agents
}
self.saveDict(root)
self.logger.debug("agents.defaults.workspace updated set=\(!trimmed.isEmpty)")
}
static func gatewayPassword() -> String? {
let root = self.loadDict()
guard let gateway = root["gateway"] as? [String: Any],
let remote = gateway["remote"] as? [String: Any]
else {
return nil
}
return remote["password"] as? String
}
static func gatewayPort() -> Int? {
let root = self.loadDict()
guard let gateway = root["gateway"] as? [String: Any] else { return nil }
if let port = gateway["port"] as? Int, port > 0 { return port }
if let number = gateway["port"] as? NSNumber, number.intValue > 0 {
return number.intValue
}
if let raw = gateway["port"] as? String,
let parsed = Int(raw.trimmingCharacters(in: .whitespacesAndNewlines)),
parsed > 0
{
return parsed
}
return nil
}
static func remoteGatewayPort() -> Int? {
guard let url = self.remoteGatewayUrl(),
let port = url.port,
port > 0
else { return nil }
return port
}
static func remoteGatewayPort(matchingHost sshHost: String) -> Int? {
let trimmedSshHost = sshHost.trimmingCharacters(in: .whitespacesAndNewlines)
guard !trimmedSshHost.isEmpty,
let url = self.remoteGatewayUrl(),
let port = url.port,
port > 0,
let urlHost = url.host?.trimmingCharacters(in: .whitespacesAndNewlines),
!urlHost.isEmpty
else {
return nil
}
let sshKey = Self.hostKey(trimmedSshHost)
let urlKey = Self.hostKey(urlHost)
guard !sshKey.isEmpty, !urlKey.isEmpty, sshKey == urlKey else { return nil }
return port
}
static func setRemoteGatewayUrl(host: String, port: Int?) {
guard let port, port > 0 else { return }
let trimmedHost = host.trimmingCharacters(in: .whitespacesAndNewlines)
guard !trimmedHost.isEmpty else { return }
self.updateGatewayDict { gateway in
var remote = gateway["remote"] as? [String: Any] ?? [:]
let existingUrl = (remote["url"] as? String)?
.trimmingCharacters(in: .whitespacesAndNewlines) ?? ""
let scheme = URL(string: existingUrl)?.scheme ?? "ws"
remote["url"] = "\(scheme)://\(trimmedHost):\(port)"
gateway["remote"] = remote
}
}
private static func remoteGatewayUrl() -> URL? {
let root = self.loadDict()
guard let gateway = root["gateway"] as? [String: Any],
let remote = gateway["remote"] as? [String: Any],
let raw = remote["url"] as? String
else {
return nil
}
let trimmed = raw.trimmingCharacters(in: .whitespacesAndNewlines)
guard !trimmed.isEmpty, let url = URL(string: trimmed) else { return nil }
return url
}
private static func hostKey(_ host: String) -> String {
let trimmed = host.trimmingCharacters(in: .whitespacesAndNewlines).lowercased()
guard !trimmed.isEmpty else { return "" }
if trimmed.contains(":") { return trimmed }
let digits = CharacterSet(charactersIn: "0123456789.")
if trimmed.rangeOfCharacter(from: digits.inverted) == nil {
return trimmed
}
return trimmed.split(separator: ".").first.map(String.init) ?? trimmed
}
private static func parseConfigData(_ data: Data) -> [String: Any]? {
if let root = try? JSONSerialization.jsonObject(with: data) as? [String: Any] {
return root
}
let decoder = JSONDecoder()
if #available(macOS 12.0, *) {
decoder.allowsJSON5 = true
}
if let decoded = try? decoder.decode([String: AnyCodable].self, from: data) {
self.logger.notice("config parsed with JSON5 decoder")
return decoded.mapValues { $0.foundationValue }
}
return nil
}
}

View File

@@ -0,0 +1,118 @@
import CoreServices
import Foundation
final class ConfigFileWatcher: @unchecked Sendable {
private let url: URL
private let queue: DispatchQueue
private var stream: FSEventStreamRef?
private var pending = false
private let onChange: () -> Void
private let watchedDir: URL
private let targetPath: String
private let targetName: String
init(url: URL, onChange: @escaping () -> Void) {
self.url = url
self.queue = DispatchQueue(label: "bot.molt.configwatcher")
self.onChange = onChange
self.watchedDir = url.deletingLastPathComponent()
self.targetPath = url.path
self.targetName = url.lastPathComponent
}
deinit {
self.stop()
}
func start() {
guard self.stream == nil else { return }
let retainedSelf = Unmanaged.passRetained(self)
var context = FSEventStreamContext(
version: 0,
info: retainedSelf.toOpaque(),
retain: nil,
release: { pointer in
guard let pointer else { return }
Unmanaged<ConfigFileWatcher>.fromOpaque(pointer).release()
},
copyDescription: nil)
let paths = [self.watchedDir.path] as CFArray
let flags = FSEventStreamCreateFlags(
kFSEventStreamCreateFlagFileEvents |
kFSEventStreamCreateFlagUseCFTypes |
kFSEventStreamCreateFlagNoDefer)
guard let stream = FSEventStreamCreate(
kCFAllocatorDefault,
Self.callback,
&context,
paths,
FSEventStreamEventId(kFSEventStreamEventIdSinceNow),
0.05,
flags)
else {
retainedSelf.release()
return
}
self.stream = stream
FSEventStreamSetDispatchQueue(stream, self.queue)
if FSEventStreamStart(stream) == false {
self.stream = nil
FSEventStreamSetDispatchQueue(stream, nil)
FSEventStreamInvalidate(stream)
FSEventStreamRelease(stream)
}
}
func stop() {
guard let stream = self.stream else { return }
self.stream = nil
FSEventStreamStop(stream)
FSEventStreamSetDispatchQueue(stream, nil)
FSEventStreamInvalidate(stream)
FSEventStreamRelease(stream)
}
}
extension ConfigFileWatcher {
private static let callback: FSEventStreamCallback = { _, info, numEvents, eventPaths, eventFlags, _ in
guard let info else { return }
let watcher = Unmanaged<ConfigFileWatcher>.fromOpaque(info).takeUnretainedValue()
watcher.handleEvents(
numEvents: numEvents,
eventPaths: eventPaths,
eventFlags: eventFlags)
}
private func handleEvents(
numEvents: Int,
eventPaths: UnsafeMutableRawPointer?,
eventFlags: UnsafePointer<FSEventStreamEventFlags>?)
{
guard numEvents > 0 else { return }
guard eventFlags != nil else { return }
guard self.matchesTarget(eventPaths: eventPaths) else { return }
if self.pending { return }
self.pending = true
self.queue.asyncAfter(deadline: .now() + 0.12) { [weak self] in
guard let self else { return }
self.pending = false
self.onChange()
}
}
private func matchesTarget(eventPaths: UnsafeMutableRawPointer?) -> Bool {
guard let eventPaths else { return true }
let paths = unsafeBitCast(eventPaths, to: NSArray.self)
for case let path as String in paths {
if path == self.targetPath { return true }
if path.hasSuffix("/\(self.targetName)") { return true }
if path == self.watchedDir.path { return true }
}
return false
}
}

View File

@@ -0,0 +1,79 @@
import Foundation
import OSLog
@MainActor
final class ConnectionModeCoordinator {
static let shared = ConnectionModeCoordinator()
private let logger = Logger(subsystem: "bot.molt", category: "connection")
private var lastMode: AppState.ConnectionMode?
/// Apply the requested connection mode by starting/stopping local gateway,
/// managing the control-channel SSH tunnel, and cleaning up chat windows/panels.
func apply(mode: AppState.ConnectionMode, paused: Bool) async {
if let lastMode = self.lastMode, lastMode != mode {
GatewayProcessManager.shared.clearLastFailure()
NodesStore.shared.lastError = nil
}
self.lastMode = mode
switch mode {
case .unconfigured:
_ = await NodeServiceManager.stop()
NodesStore.shared.lastError = nil
await RemoteTunnelManager.shared.stopAll()
WebChatManager.shared.resetTunnels()
GatewayProcessManager.shared.stop()
await GatewayConnection.shared.shutdown()
await ControlChannel.shared.disconnect()
Task.detached { await PortGuardian.shared.sweep(mode: .unconfigured) }
case .local:
_ = await NodeServiceManager.stop()
NodesStore.shared.lastError = nil
await RemoteTunnelManager.shared.stopAll()
WebChatManager.shared.resetTunnels()
let shouldStart = GatewayAutostartPolicy.shouldStartGateway(mode: .local, paused: paused)
if shouldStart {
GatewayProcessManager.shared.setActive(true)
if GatewayAutostartPolicy.shouldEnsureLaunchAgent(
mode: .local,
paused: paused)
{
Task { await GatewayProcessManager.shared.ensureLaunchAgentEnabledIfNeeded() }
}
_ = await GatewayProcessManager.shared.waitForGatewayReady()
} else {
GatewayProcessManager.shared.stop()
}
do {
try await ControlChannel.shared.configure(mode: .local)
} catch {
// Control channel will mark itself degraded; nothing else to do here.
self.logger.error(
"control channel local configure failed: \(error.localizedDescription, privacy: .public)")
}
Task.detached { await PortGuardian.shared.sweep(mode: .local) }
case .remote:
// Never run a local gateway in remote mode.
GatewayProcessManager.shared.stop()
WebChatManager.shared.resetTunnels()
do {
NodesStore.shared.lastError = nil
if let error = await NodeServiceManager.start() {
NodesStore.shared.lastError = "Node service start failed: \(error)"
}
_ = try await GatewayEndpointStore.shared.ensureRemoteControlTunnel()
let settings = CommandResolver.connectionSettings()
try await ControlChannel.shared.configure(mode: .remote(
target: settings.target,
identity: settings.identity))
} catch {
self.logger.error("remote tunnel/configure failed: \(error.localizedDescription, privacy: .public)")
}
Task.detached { await PortGuardian.shared.sweep(mode: .remote) }
}
}
}

View File

@@ -0,0 +1,44 @@
import Foundation
let launchdLabel = "bot.molt.mac"
let gatewayLaunchdLabel = "bot.molt.gateway"
let onboardingVersionKey = "moltbot.onboardingVersion"
let currentOnboardingVersion = 7
let pauseDefaultsKey = "moltbot.pauseEnabled"
let iconAnimationsEnabledKey = "moltbot.iconAnimationsEnabled"
let swabbleEnabledKey = "moltbot.swabbleEnabled"
let swabbleTriggersKey = "moltbot.swabbleTriggers"
let voiceWakeTriggerChimeKey = "moltbot.voiceWakeTriggerChime"
let voiceWakeSendChimeKey = "moltbot.voiceWakeSendChime"
let showDockIconKey = "moltbot.showDockIcon"
let defaultVoiceWakeTriggers = ["clawd", "claude"]
let voiceWakeMaxWords = 32
let voiceWakeMaxWordLength = 64
let voiceWakeMicKey = "moltbot.voiceWakeMicID"
let voiceWakeMicNameKey = "moltbot.voiceWakeMicName"
let voiceWakeLocaleKey = "moltbot.voiceWakeLocaleID"
let voiceWakeAdditionalLocalesKey = "moltbot.voiceWakeAdditionalLocaleIDs"
let voicePushToTalkEnabledKey = "moltbot.voicePushToTalkEnabled"
let talkEnabledKey = "moltbot.talkEnabled"
let iconOverrideKey = "moltbot.iconOverride"
let connectionModeKey = "moltbot.connectionMode"
let remoteTargetKey = "moltbot.remoteTarget"
let remoteIdentityKey = "moltbot.remoteIdentity"
let remoteProjectRootKey = "moltbot.remoteProjectRoot"
let remoteCliPathKey = "moltbot.remoteCliPath"
let canvasEnabledKey = "moltbot.canvasEnabled"
let cameraEnabledKey = "moltbot.cameraEnabled"
let systemRunPolicyKey = "moltbot.systemRunPolicy"
let systemRunAllowlistKey = "moltbot.systemRunAllowlist"
let systemRunEnabledKey = "moltbot.systemRunEnabled"
let locationModeKey = "moltbot.locationMode"
let locationPreciseKey = "moltbot.locationPreciseEnabled"
let peekabooBridgeEnabledKey = "moltbot.peekabooBridgeEnabled"
let deepLinkKeyKey = "moltbot.deepLinkKey"
let modelCatalogPathKey = "moltbot.modelCatalogPath"
let modelCatalogReloadKey = "moltbot.modelCatalogReload"
let cliInstallPromptedVersionKey = "moltbot.cliInstallPromptedVersion"
let heartbeatsEnabledKey = "moltbot.heartbeatsEnabled"
let debugFileLogEnabledKey = "moltbot.debug.fileLogEnabled"
let appLogLevelKey = "moltbot.debug.appLogLevel"
let voiceWakeSupported: Bool = ProcessInfo.processInfo.operatingSystemVersion.majorVersion >= 26

View File

@@ -0,0 +1,427 @@
import MoltbotKit
import MoltbotProtocol
import Foundation
import Observation
import SwiftUI
struct ControlHeartbeatEvent: Codable {
let ts: Double
let status: String
let to: String?
let preview: String?
let durationMs: Double?
let hasMedia: Bool?
let reason: String?
}
struct ControlAgentEvent: Codable, Sendable, Identifiable {
var id: String { "\(self.runId)-\(self.seq)" }
let runId: String
let seq: Int
let stream: String
let ts: Double
let data: [String: MoltbotProtocol.AnyCodable]
let summary: String?
}
enum ControlChannelError: Error, LocalizedError {
case disconnected
case badResponse(String)
var errorDescription: String? {
switch self {
case .disconnected: "Control channel disconnected"
case let .badResponse(msg): msg
}
}
}
@MainActor
@Observable
final class ControlChannel {
static let shared = ControlChannel()
enum Mode {
case local
case remote(target: String, identity: String)
}
enum ConnectionState: Equatable {
case disconnected
case connecting
case connected
case degraded(String)
}
private(set) var state: ConnectionState = .disconnected {
didSet {
CanvasManager.shared.refreshDebugStatus()
guard oldValue != self.state else { return }
switch self.state {
case .connected:
self.logger.info("control channel state -> connected")
case .connecting:
self.logger.info("control channel state -> connecting")
case .disconnected:
self.logger.info("control channel state -> disconnected")
self.scheduleRecovery(reason: "disconnected")
case let .degraded(message):
let detail = message.isEmpty ? "degraded" : "degraded: \(message)"
self.logger.info("control channel state -> \(detail, privacy: .public)")
self.scheduleRecovery(reason: message)
}
}
}
private(set) var lastPingMs: Double?
private(set) var authSourceLabel: String?
private let logger = Logger(subsystem: "bot.molt", category: "control")
private var eventTask: Task<Void, Never>?
private var recoveryTask: Task<Void, Never>?
private var lastRecoveryAt: Date?
private init() {
self.startEventStream()
}
func configure() async {
self.logger.info("control channel configure mode=local")
await self.refreshEndpoint(reason: "configure")
}
func configure(mode: Mode = .local) async throws {
switch mode {
case .local:
await self.configure()
case let .remote(target, identity):
do {
_ = (target, identity)
let idSet = !identity.trimmingCharacters(in: .whitespacesAndNewlines).isEmpty
self.logger.info(
"control channel configure mode=remote " +
"target=\(target, privacy: .public) identitySet=\(idSet, privacy: .public)")
self.state = .connecting
_ = try await GatewayEndpointStore.shared.ensureRemoteControlTunnel()
await self.refreshEndpoint(reason: "configure")
} catch {
self.state = .degraded(error.localizedDescription)
throw error
}
}
}
func refreshEndpoint(reason: String) async {
self.logger.info("control channel refresh endpoint reason=\(reason, privacy: .public)")
self.state = .connecting
do {
try await self.establishGatewayConnection()
self.state = .connected
PresenceReporter.shared.sendImmediate(reason: "connect")
} catch {
let message = self.friendlyGatewayMessage(error)
self.state = .degraded(message)
}
}
func disconnect() async {
await GatewayConnection.shared.shutdown()
self.state = .disconnected
self.lastPingMs = nil
self.authSourceLabel = nil
}
func health(timeout: TimeInterval? = nil) async throws -> Data {
do {
let start = Date()
var params: [String: AnyHashable]?
if let timeout {
params = ["timeout": AnyHashable(Int(timeout * 1000))]
}
let timeoutMs = (timeout ?? 15) * 1000
let payload = try await self.request(method: "health", params: params, timeoutMs: timeoutMs)
let ms = Date().timeIntervalSince(start) * 1000
self.lastPingMs = ms
self.state = .connected
return payload
} catch {
let message = self.friendlyGatewayMessage(error)
self.state = .degraded(message)
throw ControlChannelError.badResponse(message)
}
}
func lastHeartbeat() async throws -> ControlHeartbeatEvent? {
let data = try await self.request(method: "last-heartbeat")
return try JSONDecoder().decode(ControlHeartbeatEvent?.self, from: data)
}
func request(
method: String,
params: [String: AnyHashable]? = nil,
timeoutMs: Double? = nil) async throws -> Data
{
do {
let rawParams = params?.reduce(into: [String: MoltbotKit.AnyCodable]()) {
$0[$1.key] = MoltbotKit.AnyCodable($1.value.base)
}
let data = try await GatewayConnection.shared.request(
method: method,
params: rawParams,
timeoutMs: timeoutMs)
self.state = .connected
return data
} catch {
let message = self.friendlyGatewayMessage(error)
self.state = .degraded(message)
throw ControlChannelError.badResponse(message)
}
}
private func friendlyGatewayMessage(_ error: Error) -> String {
// Map URLSession/WS errors into user-facing, actionable text.
if let ctrlErr = error as? ControlChannelError, let desc = ctrlErr.errorDescription {
return desc
}
// If the gateway explicitly rejects the hello (e.g., auth/token mismatch), surface it.
if let urlErr = error as? URLError,
urlErr.code == .dataNotAllowed // used for WS close 1008 auth failures
{
let reason = urlErr.failureURLString ?? urlErr.localizedDescription
let tokenKey = CommandResolver.connectionModeIsRemote()
? "gateway.remote.token"
: "gateway.auth.token"
return
"Gateway rejected token; set \(tokenKey) (or CLAWDBOT_GATEWAY_TOKEN) " +
"or clear it on the gateway. " +
"Reason: \(reason)"
}
// Common misfire: we connected to the configured localhost port but it is occupied
// by some other process (e.g. a local dev gateway or a stuck SSH forward).
// The gateway handshake returns something we can't parse, which currently
// surfaces as "hello failed (unexpected response)". Give the user a pointer
// to free the port instead of a vague message.
let nsError = error as NSError
if nsError.domain == "Gateway",
nsError.localizedDescription.contains("hello failed (unexpected response)")
{
let port = GatewayEnvironment.gatewayPort()
return """
Gateway handshake got non-gateway data on localhost:\(port).
Another process is using that port or the SSH forward failed.
Stop the local gateway/port-forward on \(port) and retry Remote mode.
"""
}
if let urlError = error as? URLError {
let port = GatewayEnvironment.gatewayPort()
switch urlError.code {
case .cancelled:
return "Gateway connection was closed; start the gateway (localhost:\(port)) and retry."
case .cannotFindHost, .cannotConnectToHost:
let isRemote = CommandResolver.connectionModeIsRemote()
if isRemote {
return """
Cannot reach gateway at localhost:\(port).
Remote mode uses an SSH tunnel—check the SSH target and that the tunnel is running.
"""
}
return "Cannot reach gateway at localhost:\(port); ensure the gateway is running."
case .networkConnectionLost:
return "Gateway connection dropped; gateway likely restarted—retry."
case .timedOut:
return "Gateway request timed out; check gateway on localhost:\(port)."
case .notConnectedToInternet:
return "No network connectivity; cannot reach gateway."
default:
break
}
}
if nsError.domain == "Gateway", nsError.code == 5 {
let port = GatewayEnvironment.gatewayPort()
return "Gateway request timed out; check the gateway process on localhost:\(port)."
}
let detail = nsError.localizedDescription.isEmpty ? "unknown gateway error" : nsError.localizedDescription
let trimmed = detail.trimmingCharacters(in: .whitespacesAndNewlines)
if trimmed.lowercased().hasPrefix("gateway error:") { return trimmed }
return "Gateway error: \(trimmed)"
}
private func scheduleRecovery(reason: String) {
let now = Date()
if let last = self.lastRecoveryAt, now.timeIntervalSince(last) < 10 { return }
guard self.recoveryTask == nil else { return }
self.lastRecoveryAt = now
self.recoveryTask = Task { [weak self] in
guard let self else { return }
let mode = await MainActor.run { AppStateStore.shared.connectionMode }
guard mode != .unconfigured else {
self.recoveryTask = nil
return
}
let trimmedReason = reason.trimmingCharacters(in: .whitespacesAndNewlines)
let reasonText = trimmedReason.isEmpty ? "unknown" : trimmedReason
self.logger.info(
"control channel recovery starting " +
"mode=\(String(describing: mode), privacy: .public) " +
"reason=\(reasonText, privacy: .public)")
if mode == .local {
GatewayProcessManager.shared.setActive(true)
}
if mode == .remote {
do {
let port = try await GatewayEndpointStore.shared.ensureRemoteControlTunnel()
self.logger.info("control channel recovery ensured SSH tunnel port=\(port, privacy: .public)")
} catch {
self.logger.error(
"control channel recovery tunnel failed \(error.localizedDescription, privacy: .public)")
}
}
await self.refreshEndpoint(reason: "recovery:\(reasonText)")
if case .connected = self.state {
self.logger.info("control channel recovery finished")
} else if case let .degraded(message) = self.state {
self.logger.error("control channel recovery failed \(message, privacy: .public)")
}
self.recoveryTask = nil
}
}
private func establishGatewayConnection(timeoutMs: Int = 5000) async throws {
try await GatewayConnection.shared.refresh()
let ok = try await GatewayConnection.shared.healthOK(timeoutMs: timeoutMs)
if ok == false {
throw NSError(
domain: "Gateway",
code: 0,
userInfo: [NSLocalizedDescriptionKey: "gateway health not ok"])
}
await self.refreshAuthSourceLabel()
}
private func refreshAuthSourceLabel() async {
let isRemote = CommandResolver.connectionModeIsRemote()
let authSource = await GatewayConnection.shared.authSource()
self.authSourceLabel = Self.formatAuthSource(authSource, isRemote: isRemote)
}
private static func formatAuthSource(_ source: GatewayAuthSource?, isRemote: Bool) -> String? {
guard let source else { return nil }
switch source {
case .deviceToken:
return "Auth: device token (paired device)"
case .sharedToken:
return "Auth: shared token (\(isRemote ? "gateway.remote.token" : "gateway.auth.token"))"
case .password:
return "Auth: password (\(isRemote ? "gateway.remote.password" : "gateway.auth.password"))"
case .none:
return "Auth: none"
}
}
func sendSystemEvent(_ text: String, params: [String: AnyHashable] = [:]) async throws {
var merged = params
merged["text"] = AnyHashable(text)
_ = try await self.request(method: "system-event", params: merged)
}
private func startEventStream() {
self.eventTask?.cancel()
self.eventTask = Task { [weak self] in
guard let self else { return }
let stream = await GatewayConnection.shared.subscribe()
for await push in stream {
if Task.isCancelled { return }
await MainActor.run { [weak self] in
self?.handle(push: push)
}
}
}
}
private func handle(push: GatewayPush) {
switch push {
case let .event(evt) where evt.event == "agent":
if let payload = evt.payload,
let agent = try? GatewayPayloadDecoding.decode(payload, as: ControlAgentEvent.self)
{
AgentEventStore.shared.append(agent)
self.routeWorkActivity(from: agent)
}
case let .event(evt) where evt.event == "heartbeat":
if let payload = evt.payload,
let heartbeat = try? GatewayPayloadDecoding.decode(payload, as: ControlHeartbeatEvent.self),
let data = try? JSONEncoder().encode(heartbeat)
{
NotificationCenter.default.post(name: .controlHeartbeat, object: data)
}
case let .event(evt) where evt.event == "shutdown":
self.state = .degraded("gateway shutdown")
case .snapshot:
self.state = .connected
default:
break
}
}
private func routeWorkActivity(from event: ControlAgentEvent) {
// We currently treat VoiceWake as the "main" session for UI purposes.
// In the future, the gateway can include a sessionKey to distinguish runs.
let sessionKey = (event.data["sessionKey"]?.value as? String) ?? "main"
switch event.stream.lowercased() {
case "job":
if let state = event.data["state"]?.value as? String {
WorkActivityStore.shared.handleJob(sessionKey: sessionKey, state: state)
}
case "tool":
let phase = event.data["phase"]?.value as? String ?? ""
let name = event.data["name"]?.value as? String
let meta = event.data["meta"]?.value as? String
let args = Self.bridgeToProtocolArgs(event.data["args"])
WorkActivityStore.shared.handleTool(
sessionKey: sessionKey,
phase: phase,
name: name,
meta: meta,
args: args)
default:
break
}
}
private static func bridgeToProtocolArgs(
_ value: MoltbotProtocol.AnyCodable?) -> [String: MoltbotProtocol.AnyCodable]?
{
guard let value else { return nil }
if let dict = value.value as? [String: MoltbotProtocol.AnyCodable] {
return dict
}
if let dict = value.value as? [String: MoltbotKit.AnyCodable],
let data = try? JSONEncoder().encode(dict),
let decoded = try? JSONDecoder().decode([String: MoltbotProtocol.AnyCodable].self, from: data)
{
return decoded
}
if let data = try? JSONEncoder().encode(value),
let decoded = try? JSONDecoder().decode([String: MoltbotProtocol.AnyCodable].self, from: data)
{
return decoded
}
return nil
}
}
extension Notification.Name {
static let controlHeartbeat = Notification.Name("moltbot.control.heartbeat")
static let controlAgentEvent = Notification.Name("moltbot.control.agent")
}

View File

@@ -0,0 +1,200 @@
import MoltbotKit
import MoltbotProtocol
import Foundation
import Observation
import OSLog
@MainActor
@Observable
final class CronJobsStore {
static let shared = CronJobsStore()
var jobs: [CronJob] = []
var selectedJobId: String?
var runEntries: [CronRunLogEntry] = []
var schedulerEnabled: Bool?
var schedulerStorePath: String?
var schedulerNextWakeAtMs: Int?
var isLoadingJobs = false
var isLoadingRuns = false
var lastError: String?
var statusMessage: String?
private let logger = Logger(subsystem: "bot.molt", category: "cron.ui")
private var refreshTask: Task<Void, Never>?
private var runsTask: Task<Void, Never>?
private var eventTask: Task<Void, Never>?
private var pollTask: Task<Void, Never>?
private let interval: TimeInterval = 30
private let isPreview: Bool
init(isPreview: Bool = ProcessInfo.processInfo.isPreview) {
self.isPreview = isPreview
}
func start() {
guard !self.isPreview else { return }
guard self.eventTask == nil else { return }
self.startGatewaySubscription()
self.pollTask = Task.detached { [weak self] in
guard let self else { return }
await self.refreshJobs()
while !Task.isCancelled {
try? await Task.sleep(nanoseconds: UInt64(self.interval * 1_000_000_000))
await self.refreshJobs()
}
}
}
func stop() {
self.refreshTask?.cancel()
self.refreshTask = nil
self.runsTask?.cancel()
self.runsTask = nil
self.eventTask?.cancel()
self.eventTask = nil
self.pollTask?.cancel()
self.pollTask = nil
}
func refreshJobs() async {
guard !self.isLoadingJobs else { return }
self.isLoadingJobs = true
self.lastError = nil
self.statusMessage = nil
defer { self.isLoadingJobs = false }
do {
if let status = try? await GatewayConnection.shared.cronStatus() {
self.schedulerEnabled = status.enabled
self.schedulerStorePath = status.storePath
self.schedulerNextWakeAtMs = status.nextWakeAtMs
}
self.jobs = try await GatewayConnection.shared.cronList(includeDisabled: true)
if self.jobs.isEmpty {
self.statusMessage = "No cron jobs yet."
}
} catch {
self.logger.error("cron.list failed \(error.localizedDescription, privacy: .public)")
self.lastError = error.localizedDescription
}
}
func refreshRuns(jobId: String, limit: Int = 200) async {
guard !self.isLoadingRuns else { return }
self.isLoadingRuns = true
defer { self.isLoadingRuns = false }
do {
self.runEntries = try await GatewayConnection.shared.cronRuns(jobId: jobId, limit: limit)
} catch {
self.logger.error("cron.runs failed \(error.localizedDescription, privacy: .public)")
self.lastError = error.localizedDescription
}
}
func runJob(id: String, force: Bool = true) async {
do {
try await GatewayConnection.shared.cronRun(jobId: id, force: force)
} catch {
self.lastError = error.localizedDescription
}
}
func removeJob(id: String) async {
do {
try await GatewayConnection.shared.cronRemove(jobId: id)
await self.refreshJobs()
if self.selectedJobId == id {
self.selectedJobId = nil
self.runEntries = []
}
} catch {
self.lastError = error.localizedDescription
}
}
func setJobEnabled(id: String, enabled: Bool) async {
do {
try await GatewayConnection.shared.cronUpdate(
jobId: id,
patch: ["enabled": AnyCodable(enabled)])
await self.refreshJobs()
} catch {
self.lastError = error.localizedDescription
}
}
func upsertJob(
id: String?,
payload: [String: AnyCodable]) async throws
{
if let id {
try await GatewayConnection.shared.cronUpdate(jobId: id, patch: payload)
} else {
try await GatewayConnection.shared.cronAdd(payload: payload)
}
await self.refreshJobs()
}
// MARK: - Gateway events
private func startGatewaySubscription() {
self.eventTask?.cancel()
self.eventTask = Task { [weak self] in
guard let self else { return }
let stream = await GatewayConnection.shared.subscribe()
for await push in stream {
if Task.isCancelled { return }
await MainActor.run { [weak self] in
self?.handle(push: push)
}
}
}
}
private func handle(push: GatewayPush) {
switch push {
case let .event(evt) where evt.event == "cron":
guard let payload = evt.payload else { return }
if let cronEvt = try? GatewayPayloadDecoding.decode(payload, as: CronEvent.self) {
self.handle(cronEvent: cronEvt)
}
case .seqGap:
self.scheduleRefresh()
default:
break
}
}
private func handle(cronEvent evt: CronEvent) {
// Keep UI in sync with the gateway scheduler.
self.scheduleRefresh(delayMs: 250)
if evt.action == "finished", let selected = self.selectedJobId, selected == evt.jobId {
self.scheduleRunsRefresh(jobId: selected, delayMs: 200)
}
}
private func scheduleRefresh(delayMs: Int = 250) {
self.refreshTask?.cancel()
self.refreshTask = Task { [weak self] in
guard let self else { return }
try? await Task.sleep(nanoseconds: UInt64(delayMs) * 1_000_000)
await self.refreshJobs()
}
}
private func scheduleRunsRefresh(jobId: String, delayMs: Int = 200) {
self.runsTask?.cancel()
self.runsTask = Task { [weak self] in
guard let self else { return }
try? await Task.sleep(nanoseconds: UInt64(delayMs) * 1_000_000)
await self.refreshRuns(jobId: jobId)
}
}
// MARK: - (no additional RPC helpers)
}

View File

@@ -0,0 +1,151 @@
import AppKit
import MoltbotKit
import Foundation
import OSLog
import Security
private let deepLinkLogger = Logger(subsystem: "bot.molt", category: "DeepLink")
@MainActor
final class DeepLinkHandler {
static let shared = DeepLinkHandler()
private var lastPromptAt: Date = .distantPast
// Ephemeral, in-memory key used for unattended deep links originating from the in-app Canvas.
// This avoids blocking Canvas init on UserDefaults and doesn't weaken the external deep-link prompt:
// outside callers can't know this randomly generated key.
private nonisolated static let canvasUnattendedKey: String = DeepLinkHandler.generateRandomKey()
func handle(url: URL) async {
guard let route = DeepLinkParser.parse(url) else {
deepLinkLogger.debug("ignored url \(url.absoluteString, privacy: .public)")
return
}
guard !AppStateStore.shared.isPaused else {
self.presentAlert(title: "Moltbot is paused", message: "Unpause Moltbot to run agent actions.")
return
}
switch route {
case let .agent(link):
await self.handleAgent(link: link, originalURL: url)
}
}
private func handleAgent(link: AgentDeepLink, originalURL: URL) async {
let messagePreview = link.message.trimmingCharacters(in: .whitespacesAndNewlines)
if messagePreview.count > 20000 {
self.presentAlert(title: "Deep link too large", message: "Message exceeds 20,000 characters.")
return
}
let allowUnattended = link.key == Self.canvasUnattendedKey || link.key == Self.expectedKey()
if !allowUnattended {
if Date().timeIntervalSince(self.lastPromptAt) < 1.0 {
deepLinkLogger.debug("throttling deep link prompt")
return
}
self.lastPromptAt = Date()
let trimmed = messagePreview.count > 240 ? "\(messagePreview.prefix(240))" : messagePreview
let body =
"Run the agent with this message?\n\n\(trimmed)\n\nURL:\n\(originalURL.absoluteString)"
guard self.confirm(title: "Run Moltbot agent?", message: body) else { return }
}
if AppStateStore.shared.connectionMode == .local {
GatewayProcessManager.shared.setActive(true)
}
do {
let channel = GatewayAgentChannel(raw: link.channel)
let explicitSessionKey = link.sessionKey?
.trimmingCharacters(in: .whitespacesAndNewlines)
.nonEmpty
let resolvedSessionKey: String = if let explicitSessionKey {
explicitSessionKey
} else {
await GatewayConnection.shared.mainSessionKey()
}
let invocation = GatewayAgentInvocation(
message: messagePreview,
sessionKey: resolvedSessionKey,
thinking: link.thinking?.trimmingCharacters(in: .whitespacesAndNewlines).nonEmpty,
deliver: channel.shouldDeliver(link.deliver),
to: link.to?.trimmingCharacters(in: .whitespacesAndNewlines).nonEmpty,
channel: channel,
timeoutSeconds: link.timeoutSeconds,
idempotencyKey: UUID().uuidString)
let res = await GatewayConnection.shared.sendAgent(invocation)
if !res.ok {
throw NSError(
domain: "DeepLink",
code: 1,
userInfo: [NSLocalizedDescriptionKey: res.error ?? "agent request failed"])
}
} catch {
self.presentAlert(title: "Agent request failed", message: error.localizedDescription)
}
}
// MARK: - Auth
static func currentKey() -> String {
self.expectedKey()
}
static func currentCanvasKey() -> String {
self.canvasUnattendedKey
}
private static func expectedKey() -> String {
let defaults = UserDefaults.standard
if let key = defaults.string(forKey: deepLinkKeyKey), !key.isEmpty {
return key
}
var bytes = [UInt8](repeating: 0, count: 32)
_ = SecRandomCopyBytes(kSecRandomDefault, bytes.count, &bytes)
let data = Data(bytes)
let key = data
.base64EncodedString()
.replacingOccurrences(of: "+", with: "-")
.replacingOccurrences(of: "/", with: "_")
.replacingOccurrences(of: "=", with: "")
defaults.set(key, forKey: deepLinkKeyKey)
return key
}
private nonisolated static func generateRandomKey() -> String {
var bytes = [UInt8](repeating: 0, count: 32)
_ = SecRandomCopyBytes(kSecRandomDefault, bytes.count, &bytes)
let data = Data(bytes)
return data
.base64EncodedString()
.replacingOccurrences(of: "+", with: "-")
.replacingOccurrences(of: "/", with: "_")
.replacingOccurrences(of: "=", with: "")
}
// MARK: - UI
private func confirm(title: String, message: String) -> Bool {
let alert = NSAlert()
alert.messageText = title
alert.informativeText = message
alert.addButton(withTitle: "Run")
alert.addButton(withTitle: "Cancel")
alert.alertStyle = .warning
return alert.runModal() == .alertFirstButtonReturn
}
private func presentAlert(title: String, message: String) {
let alert = NSAlert()
alert.messageText = title
alert.informativeText = message
alert.addButton(withTitle: "OK")
alert.alertStyle = .informational
alert.runModal()
}
}

View File

@@ -0,0 +1,334 @@
import AppKit
import MoltbotKit
import MoltbotProtocol
import Foundation
import Observation
import OSLog
@MainActor
@Observable
final class DevicePairingApprovalPrompter {
static let shared = DevicePairingApprovalPrompter()
private let logger = Logger(subsystem: "bot.molt", category: "device-pairing")
private var task: Task<Void, Never>?
private var isStopping = false
private var isPresenting = false
private var queue: [PendingRequest] = []
var pendingCount: Int = 0
var pendingRepairCount: Int = 0
private var activeAlert: NSAlert?
private var activeRequestId: String?
private var alertHostWindow: NSWindow?
private var resolvedByRequestId: Set<String> = []
private final class AlertHostWindow: NSWindow {
override var canBecomeKey: Bool { true }
override var canBecomeMain: Bool { true }
}
private struct PairingList: Codable {
let pending: [PendingRequest]
let paired: [PairedDevice]?
}
private struct PairedDevice: Codable, Equatable {
let deviceId: String
let approvedAtMs: Double?
let displayName: String?
let platform: String?
let remoteIp: String?
}
private struct PendingRequest: Codable, Equatable, Identifiable {
let requestId: String
let deviceId: String
let publicKey: String
let displayName: String?
let platform: String?
let clientId: String?
let clientMode: String?
let role: String?
let scopes: [String]?
let remoteIp: String?
let silent: Bool?
let isRepair: Bool?
let ts: Double
var id: String { self.requestId }
}
private struct PairingResolvedEvent: Codable {
let requestId: String
let deviceId: String
let decision: String
let ts: Double
}
private enum PairingResolution: String {
case approved
case rejected
}
func start() {
guard self.task == nil else { return }
self.isStopping = false
self.task = Task { [weak self] in
guard let self else { return }
_ = try? await GatewayConnection.shared.refresh()
await self.loadPendingRequestsFromGateway()
let stream = await GatewayConnection.shared.subscribe(bufferingNewest: 200)
for await push in stream {
if Task.isCancelled { return }
await MainActor.run { [weak self] in self?.handle(push: push) }
}
}
}
func stop() {
self.isStopping = true
self.endActiveAlert()
self.task?.cancel()
self.task = nil
self.queue.removeAll(keepingCapacity: false)
self.updatePendingCounts()
self.isPresenting = false
self.activeRequestId = nil
self.alertHostWindow?.orderOut(nil)
self.alertHostWindow?.close()
self.alertHostWindow = nil
self.resolvedByRequestId.removeAll(keepingCapacity: false)
}
private func loadPendingRequestsFromGateway() async {
do {
let list: PairingList = try await GatewayConnection.shared.requestDecoded(method: .devicePairList)
await self.apply(list: list)
} catch {
self.logger.error("failed to load device pairing requests: \(error.localizedDescription, privacy: .public)")
}
}
private func apply(list: PairingList) async {
self.queue = list.pending.sorted(by: { $0.ts > $1.ts })
self.updatePendingCounts()
self.presentNextIfNeeded()
}
private func updatePendingCounts() {
self.pendingCount = self.queue.count
self.pendingRepairCount = self.queue.count(where: { $0.isRepair == true })
}
private func presentNextIfNeeded() {
guard !self.isStopping else { return }
guard !self.isPresenting else { return }
guard let next = self.queue.first else { return }
self.isPresenting = true
self.presentAlert(for: next)
}
private func presentAlert(for req: PendingRequest) {
self.logger.info("presenting device pairing alert requestId=\(req.requestId, privacy: .public)")
NSApp.activate(ignoringOtherApps: true)
let alert = NSAlert()
alert.alertStyle = .warning
alert.messageText = "Allow device to connect?"
alert.informativeText = Self.describe(req)
alert.addButton(withTitle: "Later")
alert.addButton(withTitle: "Approve")
alert.addButton(withTitle: "Reject")
if #available(macOS 11.0, *), alert.buttons.indices.contains(2) {
alert.buttons[2].hasDestructiveAction = true
}
self.activeAlert = alert
self.activeRequestId = req.requestId
let hostWindow = self.requireAlertHostWindow()
let sheetSize = alert.window.frame.size
if let screen = hostWindow.screen ?? NSScreen.main {
let bounds = screen.visibleFrame
let x = bounds.midX - (sheetSize.width / 2)
let sheetOriginY = bounds.midY - (sheetSize.height / 2)
let hostY = sheetOriginY + sheetSize.height - hostWindow.frame.height
hostWindow.setFrameOrigin(NSPoint(x: x, y: hostY))
} else {
hostWindow.center()
}
hostWindow.makeKeyAndOrderFront(nil)
alert.beginSheetModal(for: hostWindow) { [weak self] response in
Task { @MainActor [weak self] in
guard let self else { return }
self.activeRequestId = nil
self.activeAlert = nil
await self.handleAlertResponse(response, request: req)
hostWindow.orderOut(nil)
}
}
}
private func handleAlertResponse(_ response: NSApplication.ModalResponse, request: PendingRequest) async {
var shouldRemove = response != .alertFirstButtonReturn
defer {
if shouldRemove {
if self.queue.first == request {
self.queue.removeFirst()
} else {
self.queue.removeAll { $0 == request }
}
}
self.updatePendingCounts()
self.isPresenting = false
self.presentNextIfNeeded()
}
guard !self.isStopping else { return }
if self.resolvedByRequestId.remove(request.requestId) != nil {
return
}
switch response {
case .alertFirstButtonReturn:
shouldRemove = false
if let idx = self.queue.firstIndex(of: request) {
self.queue.remove(at: idx)
}
self.queue.append(request)
return
case .alertSecondButtonReturn:
_ = await self.approve(requestId: request.requestId)
case .alertThirdButtonReturn:
await self.reject(requestId: request.requestId)
default:
return
}
}
private func approve(requestId: String) async -> Bool {
do {
try await GatewayConnection.shared.devicePairApprove(requestId: requestId)
self.logger.info("approved device pairing requestId=\(requestId, privacy: .public)")
return true
} catch {
self.logger.error("approve failed requestId=\(requestId, privacy: .public)")
self.logger.error("approve failed: \(error.localizedDescription, privacy: .public)")
return false
}
}
private func reject(requestId: String) async {
do {
try await GatewayConnection.shared.devicePairReject(requestId: requestId)
self.logger.info("rejected device pairing requestId=\(requestId, privacy: .public)")
} catch {
self.logger.error("reject failed requestId=\(requestId, privacy: .public)")
self.logger.error("reject failed: \(error.localizedDescription, privacy: .public)")
}
}
private func endActiveAlert() {
guard let alert = self.activeAlert else { return }
if let parent = alert.window.sheetParent {
parent.endSheet(alert.window, returnCode: .abort)
}
self.activeAlert = nil
self.activeRequestId = nil
}
private func requireAlertHostWindow() -> NSWindow {
if let alertHostWindow {
return alertHostWindow
}
let window = AlertHostWindow(
contentRect: NSRect(x: 0, y: 0, width: 520, height: 1),
styleMask: [.borderless],
backing: .buffered,
defer: false)
window.title = ""
window.isReleasedWhenClosed = false
window.level = .floating
window.collectionBehavior = [.canJoinAllSpaces, .fullScreenAuxiliary]
window.isOpaque = false
window.hasShadow = false
window.backgroundColor = .clear
window.ignoresMouseEvents = true
self.alertHostWindow = window
return window
}
private func handle(push: GatewayPush) {
switch push {
case let .event(evt) where evt.event == "device.pair.requested":
guard let payload = evt.payload else { return }
do {
let req = try GatewayPayloadDecoding.decode(payload, as: PendingRequest.self)
self.enqueue(req)
} catch {
self.logger
.error("failed to decode device pairing request: \(error.localizedDescription, privacy: .public)")
}
case let .event(evt) where evt.event == "device.pair.resolved":
guard let payload = evt.payload else { return }
do {
let resolved = try GatewayPayloadDecoding.decode(payload, as: PairingResolvedEvent.self)
self.handleResolved(resolved)
} catch {
self.logger
.error(
"failed to decode device pairing resolution: \(error.localizedDescription, privacy: .public)")
}
default:
break
}
}
private func enqueue(_ req: PendingRequest) {
guard !self.queue.contains(req) else { return }
self.queue.append(req)
self.updatePendingCounts()
self.presentNextIfNeeded()
}
private func handleResolved(_ resolved: PairingResolvedEvent) {
let resolution = resolved.decision == PairingResolution.approved.rawValue ? PairingResolution
.approved : .rejected
if let activeRequestId, activeRequestId == resolved.requestId {
self.resolvedByRequestId.insert(resolved.requestId)
self.endActiveAlert()
let decision = resolution.rawValue
self.logger.info(
"device pairing resolved while active requestId=\(resolved.requestId, privacy: .public) " +
"decision=\(decision, privacy: .public)")
return
}
self.queue.removeAll { $0.requestId == resolved.requestId }
self.updatePendingCounts()
}
private static func describe(_ req: PendingRequest) -> String {
var lines: [String] = []
lines.append("Device: \(req.displayName ?? req.deviceId)")
if let platform = req.platform {
lines.append("Platform: \(platform)")
}
if let role = req.role {
lines.append("Role: \(role)")
}
if let scopes = req.scopes, !scopes.isEmpty {
lines.append("Scopes: \(scopes.joined(separator: ", "))")
}
if let remoteIp = req.remoteIp {
lines.append("IP: \(remoteIp)")
}
if req.isRepair == true {
lines.append("Repair: yes")
}
return lines.joined(separator: "\n")
}
}

View File

@@ -0,0 +1,116 @@
import AppKit
/// Central manager for Dock icon visibility.
/// Shows the Dock icon while any windows are visible, regardless of user preference.
final class DockIconManager: NSObject, @unchecked Sendable {
static let shared = DockIconManager()
private var windowsObservation: NSKeyValueObservation?
private let logger = Logger(subsystem: "bot.molt", category: "DockIconManager")
override private init() {
super.init()
self.setupObservers()
Task { @MainActor in
self.updateDockVisibility()
}
}
deinit {
self.windowsObservation?.invalidate()
NotificationCenter.default.removeObserver(self)
}
func updateDockVisibility() {
Task { @MainActor in
guard NSApp != nil else {
self.logger.warning("NSApp not ready, skipping Dock visibility update")
return
}
let userWantsDockHidden = !UserDefaults.standard.bool(forKey: showDockIconKey)
let visibleWindows = NSApp?.windows.filter { window in
window.isVisible &&
window.frame.width > 1 &&
window.frame.height > 1 &&
!window.isKind(of: NSPanel.self) &&
"\(type(of: window))" != "NSPopupMenuWindow" &&
window.contentViewController != nil
} ?? []
let hasVisibleWindows = !visibleWindows.isEmpty
if !userWantsDockHidden || hasVisibleWindows {
NSApp?.setActivationPolicy(.regular)
} else {
NSApp?.setActivationPolicy(.accessory)
}
}
}
func temporarilyShowDock() {
Task { @MainActor in
guard NSApp != nil else {
self.logger.warning("NSApp not ready, cannot show Dock icon")
return
}
NSApp.setActivationPolicy(.regular)
}
}
private func setupObservers() {
Task { @MainActor in
guard let app = NSApp else {
self.logger.warning("NSApp not ready, delaying Dock observers")
try? await Task.sleep(for: .milliseconds(200))
self.setupObservers()
return
}
self.windowsObservation = app.observe(\.windows, options: [.new]) { [weak self] _, _ in
Task { @MainActor in
try? await Task.sleep(for: .milliseconds(50))
self?.updateDockVisibility()
}
}
NotificationCenter.default.addObserver(
self,
selector: #selector(self.windowVisibilityChanged),
name: NSWindow.didBecomeKeyNotification,
object: nil)
NotificationCenter.default.addObserver(
self,
selector: #selector(self.windowVisibilityChanged),
name: NSWindow.didResignKeyNotification,
object: nil)
NotificationCenter.default.addObserver(
self,
selector: #selector(self.windowVisibilityChanged),
name: NSWindow.willCloseNotification,
object: nil)
NotificationCenter.default.addObserver(
self,
selector: #selector(self.dockPreferenceChanged),
name: UserDefaults.didChangeNotification,
object: nil)
}
}
@objc
private func windowVisibilityChanged(_: Notification) {
Task { @MainActor in
self.updateDockVisibility()
}
}
@objc
private func dockPreferenceChanged(_ notification: Notification) {
guard let userDefaults = notification.object as? UserDefaults,
userDefaults == UserDefaults.standard
else { return }
Task { @MainActor in
self.updateDockVisibility()
}
}
}

View File

@@ -0,0 +1,790 @@
import CryptoKit
import Foundation
import OSLog
import Security
enum ExecSecurity: String, CaseIterable, Codable, Identifiable {
case deny
case allowlist
case full
var id: String { self.rawValue }
var title: String {
switch self {
case .deny: "Deny"
case .allowlist: "Allowlist"
case .full: "Always Allow"
}
}
}
enum ExecApprovalQuickMode: String, CaseIterable, Identifiable {
case deny
case ask
case allow
var id: String { self.rawValue }
var title: String {
switch self {
case .deny: "Deny"
case .ask: "Always Ask"
case .allow: "Always Allow"
}
}
var security: ExecSecurity {
switch self {
case .deny: .deny
case .ask: .allowlist
case .allow: .full
}
}
var ask: ExecAsk {
switch self {
case .deny: .off
case .ask: .onMiss
case .allow: .off
}
}
static func from(security: ExecSecurity, ask: ExecAsk) -> ExecApprovalQuickMode {
switch security {
case .deny:
.deny
case .full:
.allow
case .allowlist:
.ask
}
}
}
enum ExecAsk: String, CaseIterable, Codable, Identifiable {
case off
case onMiss = "on-miss"
case always
var id: String { self.rawValue }
var title: String {
switch self {
case .off: "Never Ask"
case .onMiss: "Ask on Allowlist Miss"
case .always: "Always Ask"
}
}
}
enum ExecApprovalDecision: String, Codable, Sendable {
case allowOnce = "allow-once"
case allowAlways = "allow-always"
case deny
}
struct ExecAllowlistEntry: Codable, Hashable, Identifiable {
var id: UUID
var pattern: String
var lastUsedAt: Double?
var lastUsedCommand: String?
var lastResolvedPath: String?
init(
id: UUID = UUID(),
pattern: String,
lastUsedAt: Double? = nil,
lastUsedCommand: String? = nil,
lastResolvedPath: String? = nil)
{
self.id = id
self.pattern = pattern
self.lastUsedAt = lastUsedAt
self.lastUsedCommand = lastUsedCommand
self.lastResolvedPath = lastResolvedPath
}
private enum CodingKeys: String, CodingKey {
case id
case pattern
case lastUsedAt
case lastUsedCommand
case lastResolvedPath
}
init(from decoder: Decoder) throws {
let container = try decoder.container(keyedBy: CodingKeys.self)
self.id = try container.decodeIfPresent(UUID.self, forKey: .id) ?? UUID()
self.pattern = try container.decode(String.self, forKey: .pattern)
self.lastUsedAt = try container.decodeIfPresent(Double.self, forKey: .lastUsedAt)
self.lastUsedCommand = try container.decodeIfPresent(String.self, forKey: .lastUsedCommand)
self.lastResolvedPath = try container.decodeIfPresent(String.self, forKey: .lastResolvedPath)
}
func encode(to encoder: Encoder) throws {
var container = encoder.container(keyedBy: CodingKeys.self)
try container.encode(self.id, forKey: .id)
try container.encode(self.pattern, forKey: .pattern)
try container.encodeIfPresent(self.lastUsedAt, forKey: .lastUsedAt)
try container.encodeIfPresent(self.lastUsedCommand, forKey: .lastUsedCommand)
try container.encodeIfPresent(self.lastResolvedPath, forKey: .lastResolvedPath)
}
}
struct ExecApprovalsDefaults: Codable {
var security: ExecSecurity?
var ask: ExecAsk?
var askFallback: ExecSecurity?
var autoAllowSkills: Bool?
}
struct ExecApprovalsAgent: Codable {
var security: ExecSecurity?
var ask: ExecAsk?
var askFallback: ExecSecurity?
var autoAllowSkills: Bool?
var allowlist: [ExecAllowlistEntry]?
var isEmpty: Bool {
self.security == nil && self.ask == nil && self.askFallback == nil && self
.autoAllowSkills == nil && (self.allowlist?.isEmpty ?? true)
}
}
struct ExecApprovalsSocketConfig: Codable {
var path: String?
var token: String?
}
struct ExecApprovalsFile: Codable {
var version: Int
var socket: ExecApprovalsSocketConfig?
var defaults: ExecApprovalsDefaults?
var agents: [String: ExecApprovalsAgent]?
}
struct ExecApprovalsSnapshot: Codable {
var path: String
var exists: Bool
var hash: String
var file: ExecApprovalsFile
}
struct ExecApprovalsResolved {
let url: URL
let socketPath: String
let token: String
let defaults: ExecApprovalsResolvedDefaults
let agent: ExecApprovalsResolvedDefaults
let allowlist: [ExecAllowlistEntry]
var file: ExecApprovalsFile
}
struct ExecApprovalsResolvedDefaults {
var security: ExecSecurity
var ask: ExecAsk
var askFallback: ExecSecurity
var autoAllowSkills: Bool
}
enum ExecApprovalsStore {
private static let logger = Logger(subsystem: "bot.molt", category: "exec-approvals")
private static let defaultAgentId = "main"
private static let defaultSecurity: ExecSecurity = .deny
private static let defaultAsk: ExecAsk = .onMiss
private static let defaultAskFallback: ExecSecurity = .deny
private static let defaultAutoAllowSkills = false
static func fileURL() -> URL {
MoltbotPaths.stateDirURL.appendingPathComponent("exec-approvals.json")
}
static func socketPath() -> String {
MoltbotPaths.stateDirURL.appendingPathComponent("exec-approvals.sock").path
}
static func normalizeIncoming(_ file: ExecApprovalsFile) -> ExecApprovalsFile {
let socketPath = file.socket?.path?.trimmingCharacters(in: .whitespacesAndNewlines) ?? ""
let token = file.socket?.token?.trimmingCharacters(in: .whitespacesAndNewlines) ?? ""
var agents = file.agents ?? [:]
if let legacyDefault = agents["default"] {
if let main = agents[self.defaultAgentId] {
agents[self.defaultAgentId] = self.mergeAgents(current: main, legacy: legacyDefault)
} else {
agents[self.defaultAgentId] = legacyDefault
}
agents.removeValue(forKey: "default")
}
return ExecApprovalsFile(
version: 1,
socket: ExecApprovalsSocketConfig(
path: socketPath.isEmpty ? nil : socketPath,
token: token.isEmpty ? nil : token),
defaults: file.defaults,
agents: agents)
}
static func readSnapshot() -> ExecApprovalsSnapshot {
let url = self.fileURL()
guard FileManager().fileExists(atPath: url.path) else {
return ExecApprovalsSnapshot(
path: url.path,
exists: false,
hash: self.hashRaw(nil),
file: ExecApprovalsFile(version: 1, socket: nil, defaults: nil, agents: [:]))
}
let raw = try? String(contentsOf: url, encoding: .utf8)
let data = raw.flatMap { $0.data(using: .utf8) }
let decoded: ExecApprovalsFile = {
if let data, let file = try? JSONDecoder().decode(ExecApprovalsFile.self, from: data), file.version == 1 {
return file
}
return ExecApprovalsFile(version: 1, socket: nil, defaults: nil, agents: [:])
}()
return ExecApprovalsSnapshot(
path: url.path,
exists: true,
hash: self.hashRaw(raw),
file: decoded)
}
static func redactForSnapshot(_ file: ExecApprovalsFile) -> ExecApprovalsFile {
let socketPath = file.socket?.path?.trimmingCharacters(in: .whitespacesAndNewlines) ?? ""
if socketPath.isEmpty {
return ExecApprovalsFile(
version: file.version,
socket: nil,
defaults: file.defaults,
agents: file.agents)
}
return ExecApprovalsFile(
version: file.version,
socket: ExecApprovalsSocketConfig(path: socketPath, token: nil),
defaults: file.defaults,
agents: file.agents)
}
static func loadFile() -> ExecApprovalsFile {
let url = self.fileURL()
guard FileManager().fileExists(atPath: url.path) else {
return ExecApprovalsFile(version: 1, socket: nil, defaults: nil, agents: [:])
}
do {
let data = try Data(contentsOf: url)
let decoded = try JSONDecoder().decode(ExecApprovalsFile.self, from: data)
if decoded.version != 1 {
return ExecApprovalsFile(version: 1, socket: nil, defaults: nil, agents: [:])
}
return decoded
} catch {
self.logger.warning("exec approvals load failed: \(error.localizedDescription, privacy: .public)")
return ExecApprovalsFile(version: 1, socket: nil, defaults: nil, agents: [:])
}
}
static func saveFile(_ file: ExecApprovalsFile) {
do {
let encoder = JSONEncoder()
encoder.outputFormatting = [.prettyPrinted, .sortedKeys]
let data = try encoder.encode(file)
let url = self.fileURL()
try FileManager().createDirectory(
at: url.deletingLastPathComponent(),
withIntermediateDirectories: true)
try data.write(to: url, options: [.atomic])
try? FileManager().setAttributes([.posixPermissions: 0o600], ofItemAtPath: url.path)
} catch {
self.logger.error("exec approvals save failed: \(error.localizedDescription, privacy: .public)")
}
}
static func ensureFile() -> ExecApprovalsFile {
var file = self.loadFile()
if file.socket == nil { file.socket = ExecApprovalsSocketConfig(path: nil, token: nil) }
let path = file.socket?.path?.trimmingCharacters(in: .whitespacesAndNewlines) ?? ""
if path.isEmpty {
file.socket?.path = self.socketPath()
}
let token = file.socket?.token?.trimmingCharacters(in: .whitespacesAndNewlines) ?? ""
if token.isEmpty {
file.socket?.token = self.generateToken()
}
if file.agents == nil { file.agents = [:] }
self.saveFile(file)
return file
}
static func resolve(agentId: String?) -> ExecApprovalsResolved {
let file = self.ensureFile()
let defaults = file.defaults ?? ExecApprovalsDefaults()
let resolvedDefaults = ExecApprovalsResolvedDefaults(
security: defaults.security ?? self.defaultSecurity,
ask: defaults.ask ?? self.defaultAsk,
askFallback: defaults.askFallback ?? self.defaultAskFallback,
autoAllowSkills: defaults.autoAllowSkills ?? self.defaultAutoAllowSkills)
let key = self.agentKey(agentId)
let agentEntry = file.agents?[key] ?? ExecApprovalsAgent()
let wildcardEntry = file.agents?["*"] ?? ExecApprovalsAgent()
let resolvedAgent = ExecApprovalsResolvedDefaults(
security: agentEntry.security ?? wildcardEntry.security ?? resolvedDefaults.security,
ask: agentEntry.ask ?? wildcardEntry.ask ?? resolvedDefaults.ask,
askFallback: agentEntry.askFallback ?? wildcardEntry.askFallback
?? resolvedDefaults.askFallback,
autoAllowSkills: agentEntry.autoAllowSkills ?? wildcardEntry.autoAllowSkills
?? resolvedDefaults.autoAllowSkills)
let allowlist = ((wildcardEntry.allowlist ?? []) + (agentEntry.allowlist ?? []))
.map { entry in
ExecAllowlistEntry(
id: entry.id,
pattern: entry.pattern.trimmingCharacters(in: .whitespacesAndNewlines),
lastUsedAt: entry.lastUsedAt,
lastUsedCommand: entry.lastUsedCommand,
lastResolvedPath: entry.lastResolvedPath)
}
.filter { !$0.pattern.isEmpty }
let socketPath = self.expandPath(file.socket?.path ?? self.socketPath())
let token = file.socket?.token ?? ""
return ExecApprovalsResolved(
url: self.fileURL(),
socketPath: socketPath,
token: token,
defaults: resolvedDefaults,
agent: resolvedAgent,
allowlist: allowlist,
file: file)
}
static func resolveDefaults() -> ExecApprovalsResolvedDefaults {
let file = self.ensureFile()
let defaults = file.defaults ?? ExecApprovalsDefaults()
return ExecApprovalsResolvedDefaults(
security: defaults.security ?? self.defaultSecurity,
ask: defaults.ask ?? self.defaultAsk,
askFallback: defaults.askFallback ?? self.defaultAskFallback,
autoAllowSkills: defaults.autoAllowSkills ?? self.defaultAutoAllowSkills)
}
static func saveDefaults(_ defaults: ExecApprovalsDefaults) {
self.updateFile { file in
file.defaults = defaults
}
}
static func updateDefaults(_ mutate: (inout ExecApprovalsDefaults) -> Void) {
self.updateFile { file in
var defaults = file.defaults ?? ExecApprovalsDefaults()
mutate(&defaults)
file.defaults = defaults
}
}
static func saveAgent(_ agent: ExecApprovalsAgent, agentId: String?) {
self.updateFile { file in
var agents = file.agents ?? [:]
let key = self.agentKey(agentId)
if agent.isEmpty {
agents.removeValue(forKey: key)
} else {
agents[key] = agent
}
file.agents = agents.isEmpty ? nil : agents
}
}
static func addAllowlistEntry(agentId: String?, pattern: String) {
let trimmed = pattern.trimmingCharacters(in: .whitespacesAndNewlines)
guard !trimmed.isEmpty else { return }
self.updateFile { file in
let key = self.agentKey(agentId)
var agents = file.agents ?? [:]
var entry = agents[key] ?? ExecApprovalsAgent()
var allowlist = entry.allowlist ?? []
if allowlist.contains(where: { $0.pattern == trimmed }) { return }
allowlist.append(ExecAllowlistEntry(pattern: trimmed, lastUsedAt: Date().timeIntervalSince1970 * 1000))
entry.allowlist = allowlist
agents[key] = entry
file.agents = agents
}
}
static func recordAllowlistUse(
agentId: String?,
pattern: String,
command: String,
resolvedPath: String?)
{
self.updateFile { file in
let key = self.agentKey(agentId)
var agents = file.agents ?? [:]
var entry = agents[key] ?? ExecApprovalsAgent()
let allowlist = (entry.allowlist ?? []).map { item -> ExecAllowlistEntry in
guard item.pattern == pattern else { return item }
return ExecAllowlistEntry(
id: item.id,
pattern: item.pattern,
lastUsedAt: Date().timeIntervalSince1970 * 1000,
lastUsedCommand: command,
lastResolvedPath: resolvedPath)
}
entry.allowlist = allowlist
agents[key] = entry
file.agents = agents
}
}
static func updateAllowlist(agentId: String?, allowlist: [ExecAllowlistEntry]) {
self.updateFile { file in
let key = self.agentKey(agentId)
var agents = file.agents ?? [:]
var entry = agents[key] ?? ExecApprovalsAgent()
let cleaned = allowlist
.map { item in
ExecAllowlistEntry(
id: item.id,
pattern: item.pattern.trimmingCharacters(in: .whitespacesAndNewlines),
lastUsedAt: item.lastUsedAt,
lastUsedCommand: item.lastUsedCommand,
lastResolvedPath: item.lastResolvedPath)
}
.filter { !$0.pattern.isEmpty }
entry.allowlist = cleaned
agents[key] = entry
file.agents = agents
}
}
static func updateAgentSettings(agentId: String?, mutate: (inout ExecApprovalsAgent) -> Void) {
self.updateFile { file in
let key = self.agentKey(agentId)
var agents = file.agents ?? [:]
var entry = agents[key] ?? ExecApprovalsAgent()
mutate(&entry)
if entry.isEmpty {
agents.removeValue(forKey: key)
} else {
agents[key] = entry
}
file.agents = agents.isEmpty ? nil : agents
}
}
private static func updateFile(_ mutate: (inout ExecApprovalsFile) -> Void) {
var file = self.ensureFile()
mutate(&file)
self.saveFile(file)
}
private static func generateToken() -> String {
var bytes = [UInt8](repeating: 0, count: 24)
let status = SecRandomCopyBytes(kSecRandomDefault, bytes.count, &bytes)
if status == errSecSuccess {
return Data(bytes)
.base64EncodedString()
.replacingOccurrences(of: "+", with: "-")
.replacingOccurrences(of: "/", with: "_")
.replacingOccurrences(of: "=", with: "")
}
return UUID().uuidString
}
private static func hashRaw(_ raw: String?) -> String {
let data = Data((raw ?? "").utf8)
let digest = SHA256.hash(data: data)
return digest.map { String(format: "%02x", $0) }.joined()
}
private static func expandPath(_ raw: String) -> String {
let trimmed = raw.trimmingCharacters(in: .whitespacesAndNewlines)
if trimmed == "~" {
return FileManager().homeDirectoryForCurrentUser.path
}
if trimmed.hasPrefix("~/") {
let suffix = trimmed.dropFirst(2)
return FileManager().homeDirectoryForCurrentUser
.appendingPathComponent(String(suffix)).path
}
return trimmed
}
private static func agentKey(_ agentId: String?) -> String {
let trimmed = agentId?.trimmingCharacters(in: .whitespacesAndNewlines) ?? ""
return trimmed.isEmpty ? self.defaultAgentId : trimmed
}
private static func normalizedPattern(_ pattern: String?) -> String? {
let trimmed = pattern?.trimmingCharacters(in: .whitespacesAndNewlines) ?? ""
return trimmed.isEmpty ? nil : trimmed.lowercased()
}
private static func mergeAgents(
current: ExecApprovalsAgent,
legacy: ExecApprovalsAgent) -> ExecApprovalsAgent
{
var seen = Set<String>()
var allowlist: [ExecAllowlistEntry] = []
func append(_ entry: ExecAllowlistEntry) {
guard let key = self.normalizedPattern(entry.pattern), !seen.contains(key) else {
return
}
seen.insert(key)
allowlist.append(entry)
}
for entry in current.allowlist ?? [] {
append(entry)
}
for entry in legacy.allowlist ?? [] {
append(entry)
}
return ExecApprovalsAgent(
security: current.security ?? legacy.security,
ask: current.ask ?? legacy.ask,
askFallback: current.askFallback ?? legacy.askFallback,
autoAllowSkills: current.autoAllowSkills ?? legacy.autoAllowSkills,
allowlist: allowlist.isEmpty ? nil : allowlist)
}
}
struct ExecCommandResolution: Sendable {
let rawExecutable: String
let resolvedPath: String?
let executableName: String
let cwd: String?
static func resolve(
command: [String],
rawCommand: String?,
cwd: String?,
env: [String: String]?) -> ExecCommandResolution?
{
let trimmedRaw = rawCommand?.trimmingCharacters(in: .whitespacesAndNewlines) ?? ""
if !trimmedRaw.isEmpty, let token = self.parseFirstToken(trimmedRaw) {
return self.resolveExecutable(rawExecutable: token, cwd: cwd, env: env)
}
return self.resolve(command: command, cwd: cwd, env: env)
}
static func resolve(command: [String], cwd: String?, env: [String: String]?) -> ExecCommandResolution? {
guard let raw = command.first?.trimmingCharacters(in: .whitespacesAndNewlines), !raw.isEmpty else {
return nil
}
return self.resolveExecutable(rawExecutable: raw, cwd: cwd, env: env)
}
private static func resolveExecutable(
rawExecutable: String,
cwd: String?,
env: [String: String]?) -> ExecCommandResolution?
{
let expanded = rawExecutable.hasPrefix("~") ? (rawExecutable as NSString).expandingTildeInPath : rawExecutable
let hasPathSeparator = expanded.contains("/") || expanded.contains("\\")
let resolvedPath: String? = {
if hasPathSeparator {
if expanded.hasPrefix("/") {
return expanded
}
let base = cwd?.trimmingCharacters(in: .whitespacesAndNewlines)
let root = (base?.isEmpty == false) ? base! : FileManager().currentDirectoryPath
return URL(fileURLWithPath: root).appendingPathComponent(expanded).path
}
let searchPaths = self.searchPaths(from: env)
return CommandResolver.findExecutable(named: expanded, searchPaths: searchPaths)
}()
let name = resolvedPath.map { URL(fileURLWithPath: $0).lastPathComponent } ?? expanded
return ExecCommandResolution(
rawExecutable: expanded,
resolvedPath: resolvedPath,
executableName: name,
cwd: cwd)
}
private static func parseFirstToken(_ command: String) -> String? {
let trimmed = command.trimmingCharacters(in: .whitespacesAndNewlines)
guard !trimmed.isEmpty else { return nil }
guard let first = trimmed.first else { return nil }
if first == "\"" || first == "'" {
let rest = trimmed.dropFirst()
if let end = rest.firstIndex(of: first) {
return String(rest[..<end])
}
return String(rest)
}
return trimmed.split(whereSeparator: { $0.isWhitespace }).first.map(String.init)
}
private static func searchPaths(from env: [String: String]?) -> [String] {
let raw = env?["PATH"]
if let raw, !raw.isEmpty {
return raw.split(separator: ":").map(String.init)
}
return CommandResolver.preferredPaths()
}
}
enum ExecCommandFormatter {
static func displayString(for argv: [String]) -> String {
argv.map { arg in
let trimmed = arg.trimmingCharacters(in: .whitespacesAndNewlines)
guard !trimmed.isEmpty else { return "\"\"" }
let needsQuotes = trimmed.contains { $0.isWhitespace || $0 == "\"" }
if !needsQuotes { return trimmed }
let escaped = trimmed.replacingOccurrences(of: "\"", with: "\\\"")
return "\"\(escaped)\""
}.joined(separator: " ")
}
static func displayString(for argv: [String], rawCommand: String?) -> String {
let trimmed = rawCommand?.trimmingCharacters(in: .whitespacesAndNewlines) ?? ""
if !trimmed.isEmpty { return trimmed }
return self.displayString(for: argv)
}
}
enum ExecApprovalHelpers {
static func parseDecision(_ raw: String?) -> ExecApprovalDecision? {
let trimmed = raw?.trimmingCharacters(in: .whitespacesAndNewlines) ?? ""
guard !trimmed.isEmpty else { return nil }
return ExecApprovalDecision(rawValue: trimmed)
}
static func requiresAsk(
ask: ExecAsk,
security: ExecSecurity,
allowlistMatch: ExecAllowlistEntry?,
skillAllow: Bool) -> Bool
{
if ask == .always { return true }
if ask == .onMiss, security == .allowlist, allowlistMatch == nil, !skillAllow { return true }
return false
}
static func allowlistPattern(command: [String], resolution: ExecCommandResolution?) -> String? {
let pattern = resolution?.resolvedPath ?? resolution?.rawExecutable ?? command.first ?? ""
return pattern.isEmpty ? nil : pattern
}
}
enum ExecAllowlistMatcher {
static func match(entries: [ExecAllowlistEntry], resolution: ExecCommandResolution?) -> ExecAllowlistEntry? {
guard let resolution, !entries.isEmpty else { return nil }
let rawExecutable = resolution.rawExecutable
let resolvedPath = resolution.resolvedPath
let executableName = resolution.executableName
for entry in entries {
let pattern = entry.pattern.trimmingCharacters(in: .whitespacesAndNewlines)
if pattern.isEmpty { continue }
let hasPath = pattern.contains("/") || pattern.contains("~") || pattern.contains("\\")
if hasPath {
let target = resolvedPath ?? rawExecutable
if self.matches(pattern: pattern, target: target) { return entry }
} else if self.matches(pattern: pattern, target: executableName) {
return entry
}
}
return nil
}
private static func matches(pattern: String, target: String) -> Bool {
let trimmed = pattern.trimmingCharacters(in: .whitespacesAndNewlines)
guard !trimmed.isEmpty else { return false }
let expanded = trimmed.hasPrefix("~") ? (trimmed as NSString).expandingTildeInPath : trimmed
let normalizedPattern = self.normalizeMatchTarget(expanded)
let normalizedTarget = self.normalizeMatchTarget(target)
guard let regex = self.regex(for: normalizedPattern) else { return false }
let range = NSRange(location: 0, length: normalizedTarget.utf16.count)
return regex.firstMatch(in: normalizedTarget, options: [], range: range) != nil
}
private static func normalizeMatchTarget(_ value: String) -> String {
value.replacingOccurrences(of: "\\\\", with: "/").lowercased()
}
private static func regex(for pattern: String) -> NSRegularExpression? {
var regex = "^"
var idx = pattern.startIndex
while idx < pattern.endIndex {
let ch = pattern[idx]
if ch == "*" {
let next = pattern.index(after: idx)
if next < pattern.endIndex, pattern[next] == "*" {
regex += ".*"
idx = pattern.index(after: next)
} else {
regex += "[^/]*"
idx = next
}
continue
}
if ch == "?" {
regex += "."
idx = pattern.index(after: idx)
continue
}
regex += NSRegularExpression.escapedPattern(for: String(ch))
idx = pattern.index(after: idx)
}
regex += "$"
return try? NSRegularExpression(pattern: regex, options: [.caseInsensitive])
}
}
struct ExecEventPayload: Codable, Sendable {
var sessionKey: String
var runId: String
var host: String
var command: String?
var exitCode: Int?
var timedOut: Bool?
var success: Bool?
var output: String?
var reason: String?
static func truncateOutput(_ raw: String, maxChars: Int = 20000) -> String? {
let trimmed = raw.trimmingCharacters(in: .whitespacesAndNewlines)
guard !trimmed.isEmpty else { return nil }
if trimmed.count <= maxChars { return trimmed }
let suffix = trimmed.suffix(maxChars)
return "... (truncated) \(suffix)"
}
}
actor SkillBinsCache {
static let shared = SkillBinsCache()
private var bins: Set<String> = []
private var lastRefresh: Date?
private let refreshInterval: TimeInterval = 90
func currentBins(force: Bool = false) async -> Set<String> {
if force || self.isStale() {
await self.refresh()
}
return self.bins
}
func refresh() async {
do {
let report = try await GatewayConnection.shared.skillsStatus()
var next = Set<String>()
for skill in report.skills {
for bin in skill.requirements.bins {
let trimmed = bin.trimmingCharacters(in: .whitespacesAndNewlines)
if !trimmed.isEmpty { next.insert(trimmed) }
}
}
self.bins = next
self.lastRefresh = Date()
} catch {
if self.lastRefresh == nil {
self.bins = []
}
}
}
private func isStale() -> Bool {
guard let lastRefresh else { return true }
return Date().timeIntervalSince(lastRefresh) > self.refreshInterval
}
}

View File

@@ -0,0 +1,123 @@
import MoltbotKit
import MoltbotProtocol
import CoreGraphics
import Foundation
import OSLog
@MainActor
final class ExecApprovalsGatewayPrompter {
static let shared = ExecApprovalsGatewayPrompter()
private let logger = Logger(subsystem: "bot.molt", category: "exec-approvals.gateway")
private var task: Task<Void, Never>?
struct GatewayApprovalRequest: Codable, Sendable {
var id: String
var request: ExecApprovalPromptRequest
var createdAtMs: Int
var expiresAtMs: Int
}
func start() {
guard self.task == nil else { return }
self.task = Task { [weak self] in
await self?.run()
}
}
func stop() {
self.task?.cancel()
self.task = nil
}
private func run() async {
let stream = await GatewayConnection.shared.subscribe(bufferingNewest: 200)
for await push in stream {
if Task.isCancelled { return }
await self.handle(push: push)
}
}
private func handle(push: GatewayPush) async {
guard case let .event(evt) = push else { return }
guard evt.event == "exec.approval.requested" else { return }
guard let payload = evt.payload else { return }
do {
let data = try JSONEncoder().encode(payload)
let request = try JSONDecoder().decode(GatewayApprovalRequest.self, from: data)
guard self.shouldPresent(request: request) else { return }
let decision = ExecApprovalsPromptPresenter.prompt(request.request)
try await GatewayConnection.shared.requestVoid(
method: .execApprovalResolve,
params: [
"id": AnyCodable(request.id),
"decision": AnyCodable(decision.rawValue),
],
timeoutMs: 10000)
} catch {
self.logger.error("exec approval handling failed \(error.localizedDescription, privacy: .public)")
}
}
private func shouldPresent(request: GatewayApprovalRequest) -> Bool {
let mode = AppStateStore.shared.connectionMode
let activeSession = WebChatManager.shared.activeSessionKey?.trimmingCharacters(in: .whitespacesAndNewlines)
let requestSession = request.request.sessionKey?.trimmingCharacters(in: .whitespacesAndNewlines)
return Self.shouldPresent(
mode: mode,
activeSession: activeSession,
requestSession: requestSession,
lastInputSeconds: Self.lastInputSeconds(),
thresholdSeconds: 120)
}
private static func shouldPresent(
mode: AppState.ConnectionMode,
activeSession: String?,
requestSession: String?,
lastInputSeconds: Int?,
thresholdSeconds: Int) -> Bool
{
let active = activeSession?.trimmingCharacters(in: .whitespacesAndNewlines)
let requested = requestSession?.trimmingCharacters(in: .whitespacesAndNewlines)
let recentlyActive = lastInputSeconds.map { $0 <= thresholdSeconds } ?? (mode == .local)
if let session = requested, !session.isEmpty {
if let active, !active.isEmpty {
return active == session
}
return recentlyActive
}
if let active, !active.isEmpty {
return true
}
return mode == .local
}
private static func lastInputSeconds() -> Int? {
let anyEvent = CGEventType(rawValue: UInt32.max) ?? .null
let seconds = CGEventSource.secondsSinceLastEventType(.combinedSessionState, eventType: anyEvent)
if seconds.isNaN || seconds.isInfinite || seconds < 0 { return nil }
return Int(seconds.rounded())
}
}
#if DEBUG
extension ExecApprovalsGatewayPrompter {
static func _testShouldPresent(
mode: AppState.ConnectionMode,
activeSession: String?,
requestSession: String?,
lastInputSeconds: Int?,
thresholdSeconds: Int = 120) -> Bool
{
self.shouldPresent(
mode: mode,
activeSession: activeSession,
requestSession: requestSession,
lastInputSeconds: lastInputSeconds,
thresholdSeconds: thresholdSeconds)
}
}
#endif

View File

@@ -0,0 +1,831 @@
import AppKit
import MoltbotKit
import CryptoKit
import Darwin
import Foundation
import OSLog
struct ExecApprovalPromptRequest: Codable, Sendable {
var command: String
var cwd: String?
var host: String?
var security: String?
var ask: String?
var agentId: String?
var resolvedPath: String?
var sessionKey: String?
}
private struct ExecApprovalSocketRequest: Codable {
var type: String
var token: String
var id: String
var request: ExecApprovalPromptRequest
}
private struct ExecApprovalSocketDecision: Codable {
var type: String
var id: String
var decision: ExecApprovalDecision
}
private struct ExecHostSocketRequest: Codable {
var type: String
var id: String
var nonce: String
var ts: Int
var hmac: String
var requestJson: String
}
private struct ExecHostRequest: Codable {
var command: [String]
var rawCommand: String?
var cwd: String?
var env: [String: String]?
var timeoutMs: Int?
var needsScreenRecording: Bool?
var agentId: String?
var sessionKey: String?
var approvalDecision: ExecApprovalDecision?
}
private struct ExecHostRunResult: Codable {
var exitCode: Int?
var timedOut: Bool
var success: Bool
var stdout: String
var stderr: String
var error: String?
}
private struct ExecHostError: Codable {
var code: String
var message: String
var reason: String?
}
private struct ExecHostResponse: Codable {
var type: String
var id: String
var ok: Bool
var payload: ExecHostRunResult?
var error: ExecHostError?
}
enum ExecApprovalsSocketClient {
private struct TimeoutError: LocalizedError {
var message: String
var errorDescription: String? { self.message }
}
static func requestDecision(
socketPath: String,
token: String,
request: ExecApprovalPromptRequest,
timeoutMs: Int = 15000) async -> ExecApprovalDecision?
{
let trimmedPath = socketPath.trimmingCharacters(in: .whitespacesAndNewlines)
let trimmedToken = token.trimmingCharacters(in: .whitespacesAndNewlines)
guard !trimmedPath.isEmpty, !trimmedToken.isEmpty else { return nil }
do {
return try await AsyncTimeout.withTimeoutMs(
timeoutMs: timeoutMs,
onTimeout: {
TimeoutError(message: "exec approvals socket timeout")
},
operation: {
try await Task.detached {
try self.requestDecisionSync(
socketPath: trimmedPath,
token: trimmedToken,
request: request)
}.value
})
} catch {
return nil
}
}
private static func requestDecisionSync(
socketPath: String,
token: String,
request: ExecApprovalPromptRequest) throws -> ExecApprovalDecision?
{
let fd = socket(AF_UNIX, SOCK_STREAM, 0)
guard fd >= 0 else {
throw NSError(domain: "ExecApprovals", code: 1, userInfo: [
NSLocalizedDescriptionKey: "socket create failed",
])
}
var addr = sockaddr_un()
addr.sun_family = sa_family_t(AF_UNIX)
let maxLen = MemoryLayout.size(ofValue: addr.sun_path)
if socketPath.utf8.count >= maxLen {
throw NSError(domain: "ExecApprovals", code: 2, userInfo: [
NSLocalizedDescriptionKey: "socket path too long",
])
}
socketPath.withCString { cstr in
withUnsafeMutablePointer(to: &addr.sun_path) { ptr in
let raw = UnsafeMutableRawPointer(ptr).assumingMemoryBound(to: Int8.self)
strncpy(raw, cstr, maxLen - 1)
}
}
let size = socklen_t(MemoryLayout.size(ofValue: addr))
let result = withUnsafePointer(to: &addr) { ptr in
ptr.withMemoryRebound(to: sockaddr.self, capacity: 1) { rebound in
connect(fd, rebound, size)
}
}
if result != 0 {
throw NSError(domain: "ExecApprovals", code: 3, userInfo: [
NSLocalizedDescriptionKey: "socket connect failed",
])
}
let handle = FileHandle(fileDescriptor: fd, closeOnDealloc: true)
let message = ExecApprovalSocketRequest(
type: "request",
token: token,
id: UUID().uuidString,
request: request)
let data = try JSONEncoder().encode(message)
var payload = data
payload.append(0x0A)
try handle.write(contentsOf: payload)
guard let line = try self.readLine(from: handle, maxBytes: 256_000),
let lineData = line.data(using: .utf8)
else { return nil }
let response = try JSONDecoder().decode(ExecApprovalSocketDecision.self, from: lineData)
return response.decision
}
private static func readLine(from handle: FileHandle, maxBytes: Int) throws -> String? {
var buffer = Data()
while buffer.count < maxBytes {
let chunk = try handle.read(upToCount: 4096) ?? Data()
if chunk.isEmpty { break }
buffer.append(chunk)
if buffer.contains(0x0A) { break }
}
guard let newlineIndex = buffer.firstIndex(of: 0x0A) else {
guard !buffer.isEmpty else { return nil }
return String(data: buffer, encoding: .utf8)
}
let lineData = buffer.subdata(in: 0..<newlineIndex)
return String(data: lineData, encoding: .utf8)
}
}
@MainActor
final class ExecApprovalsPromptServer {
static let shared = ExecApprovalsPromptServer()
private var server: ExecApprovalsSocketServer?
func start() {
guard self.server == nil else { return }
let approvals = ExecApprovalsStore.resolve(agentId: nil)
let server = ExecApprovalsSocketServer(
socketPath: approvals.socketPath,
token: approvals.token,
onPrompt: { request in
await ExecApprovalsPromptPresenter.prompt(request)
},
onExec: { request in
await ExecHostExecutor.handle(request)
})
server.start()
self.server = server
}
func stop() {
self.server?.stop()
self.server = nil
}
}
enum ExecApprovalsPromptPresenter {
@MainActor
static func prompt(_ request: ExecApprovalPromptRequest) -> ExecApprovalDecision {
NSApp.activate(ignoringOtherApps: true)
let alert = NSAlert()
alert.alertStyle = .warning
alert.messageText = "Allow this command?"
alert.informativeText = "Review the command details before allowing."
alert.accessoryView = self.buildAccessoryView(request)
alert.addButton(withTitle: "Allow Once")
alert.addButton(withTitle: "Always Allow")
alert.addButton(withTitle: "Don't Allow")
if #available(macOS 11.0, *), alert.buttons.indices.contains(2) {
alert.buttons[2].hasDestructiveAction = true
}
switch alert.runModal() {
case .alertFirstButtonReturn:
return .allowOnce
case .alertSecondButtonReturn:
return .allowAlways
default:
return .deny
}
}
@MainActor
private static func buildAccessoryView(_ request: ExecApprovalPromptRequest) -> NSView {
let stack = NSStackView()
stack.orientation = .vertical
stack.spacing = 8
stack.alignment = .leading
let commandTitle = NSTextField(labelWithString: "Command")
commandTitle.font = NSFont.boldSystemFont(ofSize: NSFont.systemFontSize)
stack.addArrangedSubview(commandTitle)
let commandText = NSTextView()
commandText.isEditable = false
commandText.isSelectable = true
commandText.drawsBackground = true
commandText.backgroundColor = NSColor.textBackgroundColor
commandText.font = NSFont.monospacedSystemFont(ofSize: NSFont.systemFontSize, weight: .regular)
commandText.string = request.command
commandText.textContainerInset = NSSize(width: 6, height: 6)
commandText.textContainer?.lineFragmentPadding = 0
commandText.textContainer?.widthTracksTextView = true
commandText.isHorizontallyResizable = false
commandText.isVerticallyResizable = false
let commandScroll = NSScrollView()
commandScroll.borderType = .lineBorder
commandScroll.hasVerticalScroller = false
commandScroll.hasHorizontalScroller = false
commandScroll.documentView = commandText
commandScroll.translatesAutoresizingMaskIntoConstraints = false
commandScroll.widthAnchor.constraint(lessThanOrEqualToConstant: 440).isActive = true
commandScroll.heightAnchor.constraint(greaterThanOrEqualToConstant: 56).isActive = true
stack.addArrangedSubview(commandScroll)
let contextTitle = NSTextField(labelWithString: "Context")
contextTitle.font = NSFont.boldSystemFont(ofSize: NSFont.systemFontSize)
stack.addArrangedSubview(contextTitle)
let contextStack = NSStackView()
contextStack.orientation = .vertical
contextStack.spacing = 4
contextStack.alignment = .leading
let trimmedCwd = request.cwd?.trimmingCharacters(in: .whitespacesAndNewlines) ?? ""
if !trimmedCwd.isEmpty {
self.addDetailRow(title: "Working directory", value: trimmedCwd, to: contextStack)
}
let trimmedAgent = request.agentId?.trimmingCharacters(in: .whitespacesAndNewlines) ?? ""
if !trimmedAgent.isEmpty {
self.addDetailRow(title: "Agent", value: trimmedAgent, to: contextStack)
}
let trimmedPath = request.resolvedPath?.trimmingCharacters(in: .whitespacesAndNewlines) ?? ""
if !trimmedPath.isEmpty {
self.addDetailRow(title: "Executable", value: trimmedPath, to: contextStack)
}
let trimmedHost = request.host?.trimmingCharacters(in: .whitespacesAndNewlines) ?? ""
if !trimmedHost.isEmpty {
self.addDetailRow(title: "Host", value: trimmedHost, to: contextStack)
}
if let security = request.security?.trimmingCharacters(in: .whitespacesAndNewlines), !security.isEmpty {
self.addDetailRow(title: "Security", value: security, to: contextStack)
}
if let ask = request.ask?.trimmingCharacters(in: .whitespacesAndNewlines), !ask.isEmpty {
self.addDetailRow(title: "Ask mode", value: ask, to: contextStack)
}
if contextStack.arrangedSubviews.isEmpty {
let empty = NSTextField(labelWithString: "No additional context provided.")
empty.textColor = NSColor.secondaryLabelColor
empty.font = NSFont.systemFont(ofSize: NSFont.smallSystemFontSize)
contextStack.addArrangedSubview(empty)
}
stack.addArrangedSubview(contextStack)
let footer = NSTextField(labelWithString: "This runs on this machine.")
footer.textColor = NSColor.secondaryLabelColor
footer.font = NSFont.systemFont(ofSize: NSFont.smallSystemFontSize)
stack.addArrangedSubview(footer)
return stack
}
@MainActor
private static func addDetailRow(title: String, value: String, to stack: NSStackView) {
let row = NSStackView()
row.orientation = .horizontal
row.spacing = 6
row.alignment = .firstBaseline
let titleLabel = NSTextField(labelWithString: "\(title):")
titleLabel.font = NSFont.systemFont(ofSize: NSFont.smallSystemFontSize, weight: .semibold)
titleLabel.textColor = NSColor.secondaryLabelColor
let valueLabel = NSTextField(labelWithString: value)
valueLabel.font = NSFont.systemFont(ofSize: NSFont.smallSystemFontSize)
valueLabel.lineBreakMode = .byTruncatingMiddle
valueLabel.setContentCompressionResistancePriority(.defaultLow, for: .horizontal)
row.addArrangedSubview(titleLabel)
row.addArrangedSubview(valueLabel)
stack.addArrangedSubview(row)
}
}
@MainActor
private enum ExecHostExecutor {
private struct ExecApprovalContext {
let command: [String]
let displayCommand: String
let trimmedAgent: String?
let approvals: ExecApprovalsResolved
let security: ExecSecurity
let ask: ExecAsk
let autoAllowSkills: Bool
let env: [String: String]?
let resolution: ExecCommandResolution?
let allowlistMatch: ExecAllowlistEntry?
let skillAllow: Bool
}
private static let blockedEnvKeys: Set<String> = [
"PATH",
"NODE_OPTIONS",
"PYTHONHOME",
"PYTHONPATH",
"PERL5LIB",
"PERL5OPT",
"RUBYOPT",
]
private static let blockedEnvPrefixes: [String] = [
"DYLD_",
"LD_",
]
static func handle(_ request: ExecHostRequest) async -> ExecHostResponse {
let command = request.command.map { $0.trimmingCharacters(in: .whitespacesAndNewlines) }
guard !command.isEmpty else {
return self.errorResponse(
code: "INVALID_REQUEST",
message: "command required",
reason: "invalid")
}
let context = await self.buildContext(request: request, command: command)
if context.security == .deny {
return self.errorResponse(
code: "UNAVAILABLE",
message: "SYSTEM_RUN_DISABLED: security=deny",
reason: "security=deny")
}
let approvalDecision = request.approvalDecision
if approvalDecision == .deny {
return self.errorResponse(
code: "UNAVAILABLE",
message: "SYSTEM_RUN_DENIED: user denied",
reason: "user-denied")
}
var approvedByAsk = approvalDecision != nil
if ExecApprovalHelpers.requiresAsk(
ask: context.ask,
security: context.security,
allowlistMatch: context.allowlistMatch,
skillAllow: context.skillAllow),
approvalDecision == nil
{
let decision = ExecApprovalsPromptPresenter.prompt(
ExecApprovalPromptRequest(
command: context.displayCommand,
cwd: request.cwd,
host: "node",
security: context.security.rawValue,
ask: context.ask.rawValue,
agentId: context.trimmedAgent,
resolvedPath: context.resolution?.resolvedPath,
sessionKey: request.sessionKey))
switch decision {
case .deny:
return self.errorResponse(
code: "UNAVAILABLE",
message: "SYSTEM_RUN_DENIED: user denied",
reason: "user-denied")
case .allowAlways:
approvedByAsk = true
self.persistAllowlistEntry(decision: decision, context: context)
case .allowOnce:
approvedByAsk = true
}
}
self.persistAllowlistEntry(decision: approvalDecision, context: context)
if context.security == .allowlist,
context.allowlistMatch == nil,
!context.skillAllow,
!approvedByAsk
{
return self.errorResponse(
code: "UNAVAILABLE",
message: "SYSTEM_RUN_DENIED: allowlist miss",
reason: "allowlist-miss")
}
if let match = context.allowlistMatch {
ExecApprovalsStore.recordAllowlistUse(
agentId: context.trimmedAgent,
pattern: match.pattern,
command: context.displayCommand,
resolvedPath: context.resolution?.resolvedPath)
}
if let errorResponse = await self.ensureScreenRecordingAccess(request.needsScreenRecording) {
return errorResponse
}
return await self.runCommand(
command: command,
cwd: request.cwd,
env: context.env,
timeoutMs: request.timeoutMs)
}
private static func buildContext(request: ExecHostRequest, command: [String]) async -> ExecApprovalContext {
let displayCommand = ExecCommandFormatter.displayString(
for: command,
rawCommand: request.rawCommand)
let agentId = request.agentId?.trimmingCharacters(in: .whitespacesAndNewlines)
let trimmedAgent = (agentId?.isEmpty == false) ? agentId : nil
let approvals = ExecApprovalsStore.resolve(agentId: trimmedAgent)
let security = approvals.agent.security
let ask = approvals.agent.ask
let autoAllowSkills = approvals.agent.autoAllowSkills
let env = self.sanitizedEnv(request.env)
let resolution = ExecCommandResolution.resolve(
command: command,
rawCommand: request.rawCommand,
cwd: request.cwd,
env: env)
let allowlistMatch = security == .allowlist
? ExecAllowlistMatcher.match(entries: approvals.allowlist, resolution: resolution)
: nil
let skillAllow: Bool
if autoAllowSkills, let name = resolution?.executableName {
let bins = await SkillBinsCache.shared.currentBins()
skillAllow = bins.contains(name)
} else {
skillAllow = false
}
return ExecApprovalContext(
command: command,
displayCommand: displayCommand,
trimmedAgent: trimmedAgent,
approvals: approvals,
security: security,
ask: ask,
autoAllowSkills: autoAllowSkills,
env: env,
resolution: resolution,
allowlistMatch: allowlistMatch,
skillAllow: skillAllow)
}
private static func persistAllowlistEntry(
decision: ExecApprovalDecision?,
context: ExecApprovalContext)
{
guard decision == .allowAlways, context.security == .allowlist else { return }
guard let pattern = ExecApprovalHelpers.allowlistPattern(
command: context.command,
resolution: context.resolution)
else {
return
}
ExecApprovalsStore.addAllowlistEntry(agentId: context.trimmedAgent, pattern: pattern)
}
private static func ensureScreenRecordingAccess(_ needsScreenRecording: Bool?) async -> ExecHostResponse? {
guard needsScreenRecording == true else { return nil }
let authorized = await PermissionManager
.status([.screenRecording])[.screenRecording] ?? false
if authorized { return nil }
return self.errorResponse(
code: "UNAVAILABLE",
message: "PERMISSION_MISSING: screenRecording",
reason: "permission:screenRecording")
}
private static func runCommand(
command: [String],
cwd: String?,
env: [String: String]?,
timeoutMs: Int?) async -> ExecHostResponse
{
let timeoutSec = timeoutMs.flatMap { Double($0) / 1000.0 }
let result = await Task.detached { () -> ShellExecutor.ShellResult in
await ShellExecutor.runDetailed(
command: command,
cwd: cwd,
env: env,
timeout: timeoutSec)
}.value
let payload = ExecHostRunResult(
exitCode: result.exitCode,
timedOut: result.timedOut,
success: result.success,
stdout: result.stdout,
stderr: result.stderr,
error: result.errorMessage)
return self.successResponse(payload)
}
private static func errorResponse(
code: String,
message: String,
reason: String?) -> ExecHostResponse
{
ExecHostResponse(
type: "exec-res",
id: UUID().uuidString,
ok: false,
payload: nil,
error: ExecHostError(code: code, message: message, reason: reason))
}
private static func successResponse(_ payload: ExecHostRunResult) -> ExecHostResponse {
ExecHostResponse(
type: "exec-res",
id: UUID().uuidString,
ok: true,
payload: payload,
error: nil)
}
private static func sanitizedEnv(_ overrides: [String: String]?) -> [String: String]? {
guard let overrides else { return nil }
var merged = ProcessInfo.processInfo.environment
for (rawKey, value) in overrides {
let key = rawKey.trimmingCharacters(in: .whitespacesAndNewlines)
guard !key.isEmpty else { continue }
let upper = key.uppercased()
if self.blockedEnvKeys.contains(upper) { continue }
if self.blockedEnvPrefixes.contains(where: { upper.hasPrefix($0) }) { continue }
merged[key] = value
}
return merged
}
}
private final class ExecApprovalsSocketServer: @unchecked Sendable {
private let logger = Logger(subsystem: "bot.molt", category: "exec-approvals.socket")
private let socketPath: String
private let token: String
private let onPrompt: @Sendable (ExecApprovalPromptRequest) async -> ExecApprovalDecision
private let onExec: @Sendable (ExecHostRequest) async -> ExecHostResponse
private var socketFD: Int32 = -1
private var acceptTask: Task<Void, Never>?
private var isRunning = false
init(
socketPath: String,
token: String,
onPrompt: @escaping @Sendable (ExecApprovalPromptRequest) async -> ExecApprovalDecision,
onExec: @escaping @Sendable (ExecHostRequest) async -> ExecHostResponse)
{
self.socketPath = socketPath
self.token = token
self.onPrompt = onPrompt
self.onExec = onExec
}
func start() {
guard !self.isRunning else { return }
self.isRunning = true
self.acceptTask = Task.detached { [weak self] in
await self?.runAcceptLoop()
}
}
func stop() {
self.isRunning = false
self.acceptTask?.cancel()
self.acceptTask = nil
if self.socketFD >= 0 {
close(self.socketFD)
self.socketFD = -1
}
if !self.socketPath.isEmpty {
unlink(self.socketPath)
}
}
private func runAcceptLoop() async {
let fd = self.openSocket()
guard fd >= 0 else {
self.isRunning = false
return
}
self.socketFD = fd
while self.isRunning {
var addr = sockaddr_un()
var len = socklen_t(MemoryLayout.size(ofValue: addr))
let client = withUnsafeMutablePointer(to: &addr) { ptr in
ptr.withMemoryRebound(to: sockaddr.self, capacity: 1) { rebound in
accept(fd, rebound, &len)
}
}
if client < 0 {
if errno == EINTR { continue }
break
}
Task.detached { [weak self] in
await self?.handleClient(fd: client)
}
}
}
private func openSocket() -> Int32 {
let fd = socket(AF_UNIX, SOCK_STREAM, 0)
guard fd >= 0 else {
self.logger.error("exec approvals socket create failed")
return -1
}
unlink(self.socketPath)
var addr = sockaddr_un()
addr.sun_family = sa_family_t(AF_UNIX)
let maxLen = MemoryLayout.size(ofValue: addr.sun_path)
if self.socketPath.utf8.count >= maxLen {
self.logger.error("exec approvals socket path too long")
close(fd)
return -1
}
self.socketPath.withCString { cstr in
withUnsafeMutablePointer(to: &addr.sun_path) { ptr in
let raw = UnsafeMutableRawPointer(ptr).assumingMemoryBound(to: Int8.self)
memset(raw, 0, maxLen)
strncpy(raw, cstr, maxLen - 1)
}
}
let size = socklen_t(MemoryLayout.size(ofValue: addr))
let result = withUnsafePointer(to: &addr) { ptr in
ptr.withMemoryRebound(to: sockaddr.self, capacity: 1) { rebound in
bind(fd, rebound, size)
}
}
if result != 0 {
self.logger.error("exec approvals socket bind failed")
close(fd)
return -1
}
if listen(fd, 16) != 0 {
self.logger.error("exec approvals socket listen failed")
close(fd)
return -1
}
chmod(self.socketPath, 0o600)
self.logger.info("exec approvals socket listening at \(self.socketPath, privacy: .public)")
return fd
}
private func handleClient(fd: Int32) async {
let handle = FileHandle(fileDescriptor: fd, closeOnDealloc: true)
do {
guard self.isAllowedPeer(fd: fd) else {
try self.sendApprovalResponse(handle: handle, id: UUID().uuidString, decision: .deny)
return
}
guard let line = try self.readLine(from: handle, maxBytes: 256_000),
let data = line.data(using: .utf8)
else {
return
}
guard
let envelope = try JSONSerialization.jsonObject(with: data) as? [String: Any],
let type = envelope["type"] as? String
else {
return
}
if type == "request" {
let request = try JSONDecoder().decode(ExecApprovalSocketRequest.self, from: data)
guard request.token == self.token else {
try self.sendApprovalResponse(handle: handle, id: request.id, decision: .deny)
return
}
let decision = await self.onPrompt(request.request)
try self.sendApprovalResponse(handle: handle, id: request.id, decision: decision)
return
}
if type == "exec" {
let request = try JSONDecoder().decode(ExecHostSocketRequest.self, from: data)
let response = await self.handleExecRequest(request)
try self.sendExecResponse(handle: handle, response: response)
return
}
} catch {
self.logger.error("exec approvals socket handling failed: \(error.localizedDescription, privacy: .public)")
}
}
private func readLine(from handle: FileHandle, maxBytes: Int) throws -> String? {
var buffer = Data()
while buffer.count < maxBytes {
let chunk = try handle.read(upToCount: 4096) ?? Data()
if chunk.isEmpty { break }
buffer.append(chunk)
if buffer.contains(0x0A) { break }
}
guard let newlineIndex = buffer.firstIndex(of: 0x0A) else {
guard !buffer.isEmpty else { return nil }
return String(data: buffer, encoding: .utf8)
}
let lineData = buffer.subdata(in: 0..<newlineIndex)
return String(data: lineData, encoding: .utf8)
}
private func sendApprovalResponse(
handle: FileHandle,
id: String,
decision: ExecApprovalDecision) throws
{
let response = ExecApprovalSocketDecision(type: "decision", id: id, decision: decision)
let data = try JSONEncoder().encode(response)
var payload = data
payload.append(0x0A)
try handle.write(contentsOf: payload)
}
private func sendExecResponse(handle: FileHandle, response: ExecHostResponse) throws {
let data = try JSONEncoder().encode(response)
var payload = data
payload.append(0x0A)
try handle.write(contentsOf: payload)
}
private func isAllowedPeer(fd: Int32) -> Bool {
var uid = uid_t(0)
var gid = gid_t(0)
if getpeereid(fd, &uid, &gid) != 0 {
return false
}
return uid == geteuid()
}
private func handleExecRequest(_ request: ExecHostSocketRequest) async -> ExecHostResponse {
let nowMs = Int(Date().timeIntervalSince1970 * 1000)
if abs(nowMs - request.ts) > 10000 {
return ExecHostResponse(
type: "exec-res",
id: request.id,
ok: false,
payload: nil,
error: ExecHostError(code: "INVALID_REQUEST", message: "expired request", reason: "ttl"))
}
let expected = self.hmacHex(nonce: request.nonce, ts: request.ts, requestJson: request.requestJson)
if expected != request.hmac {
return ExecHostResponse(
type: "exec-res",
id: request.id,
ok: false,
payload: nil,
error: ExecHostError(code: "INVALID_REQUEST", message: "invalid auth", reason: "hmac"))
}
guard let requestData = request.requestJson.data(using: .utf8),
let payload = try? JSONDecoder().decode(ExecHostRequest.self, from: requestData)
else {
return ExecHostResponse(
type: "exec-res",
id: request.id,
ok: false,
payload: nil,
error: ExecHostError(code: "INVALID_REQUEST", message: "invalid payload", reason: "json"))
}
let response = await self.onExec(payload)
return ExecHostResponse(
type: "exec-res",
id: request.id,
ok: response.ok,
payload: response.payload,
error: response.error)
}
private func hmacHex(nonce: String, ts: Int, requestJson: String) -> String {
let key = SymmetricKey(data: Data(self.token.utf8))
let message = "\(nonce):\(ts):\(requestJson)"
let mac = HMAC<SHA256>.authenticationCode(for: Data(message.utf8), using: key)
return mac.map { String(format: "%02x", $0) }.joined()
}
}

View File

@@ -0,0 +1,737 @@
import MoltbotChatUI
import MoltbotKit
import MoltbotProtocol
import Foundation
import OSLog
private let gatewayConnectionLogger = Logger(subsystem: "bot.molt", category: "gateway.connection")
enum GatewayAgentChannel: String, Codable, CaseIterable, Sendable {
case last
case whatsapp
case telegram
case discord
case googlechat
case slack
case signal
case imessage
case msteams
case bluebubbles
case webchat
init(raw: String?) {
let normalized = (raw ?? "").trimmingCharacters(in: .whitespacesAndNewlines).lowercased()
self = GatewayAgentChannel(rawValue: normalized) ?? .last
}
var isDeliverable: Bool { self != .webchat }
func shouldDeliver(_ deliver: Bool) -> Bool { deliver && self.isDeliverable }
}
struct GatewayAgentInvocation: Sendable {
var message: String
var sessionKey: String = "main"
var thinking: String?
var deliver: Bool = false
var to: String?
var channel: GatewayAgentChannel = .last
var timeoutSeconds: Int?
var idempotencyKey: String = UUID().uuidString
}
/// Single, shared Gateway websocket connection for the whole app.
///
/// This owns exactly one `GatewayChannelActor` and reuses it across all callers
/// (ControlChannel, debug actions, SwiftUI WebChat, etc.).
actor GatewayConnection {
static let shared = GatewayConnection()
typealias Config = (url: URL, token: String?, password: String?)
enum Method: String, Sendable {
case agent
case status
case setHeartbeats = "set-heartbeats"
case systemEvent = "system-event"
case health
case channelsStatus = "channels.status"
case configGet = "config.get"
case configSet = "config.set"
case configPatch = "config.patch"
case configSchema = "config.schema"
case wizardStart = "wizard.start"
case wizardNext = "wizard.next"
case wizardCancel = "wizard.cancel"
case wizardStatus = "wizard.status"
case talkMode = "talk.mode"
case webLoginStart = "web.login.start"
case webLoginWait = "web.login.wait"
case channelsLogout = "channels.logout"
case modelsList = "models.list"
case chatHistory = "chat.history"
case sessionsPreview = "sessions.preview"
case chatSend = "chat.send"
case chatAbort = "chat.abort"
case skillsStatus = "skills.status"
case skillsInstall = "skills.install"
case skillsUpdate = "skills.update"
case voicewakeGet = "voicewake.get"
case voicewakeSet = "voicewake.set"
case nodePairApprove = "node.pair.approve"
case nodePairReject = "node.pair.reject"
case devicePairList = "device.pair.list"
case devicePairApprove = "device.pair.approve"
case devicePairReject = "device.pair.reject"
case execApprovalResolve = "exec.approval.resolve"
case cronList = "cron.list"
case cronRuns = "cron.runs"
case cronRun = "cron.run"
case cronRemove = "cron.remove"
case cronUpdate = "cron.update"
case cronAdd = "cron.add"
case cronStatus = "cron.status"
}
private let configProvider: @Sendable () async throws -> Config
private let sessionBox: WebSocketSessionBox?
private let decoder = JSONDecoder()
private var client: GatewayChannelActor?
private var configuredURL: URL?
private var configuredToken: String?
private var configuredPassword: String?
private var subscribers: [UUID: AsyncStream<GatewayPush>.Continuation] = [:]
private var lastSnapshot: HelloOk?
init(
configProvider: @escaping @Sendable () async throws -> Config = GatewayConnection.defaultConfigProvider,
sessionBox: WebSocketSessionBox? = nil)
{
self.configProvider = configProvider
self.sessionBox = sessionBox
}
// MARK: - Low-level request
func request(
method: String,
params: [String: AnyCodable]?,
timeoutMs: Double? = nil) async throws -> Data
{
let cfg = try await self.configProvider()
await self.configure(url: cfg.url, token: cfg.token, password: cfg.password)
guard let client else {
throw NSError(domain: "Gateway", code: 0, userInfo: [NSLocalizedDescriptionKey: "gateway not configured"])
}
do {
return try await client.request(method: method, params: params, timeoutMs: timeoutMs)
} catch {
if error is GatewayResponseError || error is GatewayDecodingError {
throw error
}
// Auto-recover in local mode by spawning/attaching a gateway and retrying a few times.
// Canvas interactions should "just work" even if the local gateway isn't running yet.
let mode = await MainActor.run { AppStateStore.shared.connectionMode }
switch mode {
case .local:
await MainActor.run { GatewayProcessManager.shared.setActive(true) }
var lastError: Error = error
for delayMs in [150, 400, 900] {
try await Task.sleep(nanoseconds: UInt64(delayMs) * 1_000_000)
do {
return try await client.request(method: method, params: params, timeoutMs: timeoutMs)
} catch {
lastError = error
}
}
let nsError = lastError as NSError
if nsError.domain == URLError.errorDomain,
let fallback = await GatewayEndpointStore.shared.maybeFallbackToTailnet(from: cfg.url)
{
await self.configure(url: fallback.url, token: fallback.token, password: fallback.password)
for delayMs in [150, 400, 900] {
try await Task.sleep(nanoseconds: UInt64(delayMs) * 1_000_000)
do {
guard let client = self.client else {
throw NSError(
domain: "Gateway",
code: 0,
userInfo: [NSLocalizedDescriptionKey: "gateway not configured"])
}
return try await client.request(method: method, params: params, timeoutMs: timeoutMs)
} catch {
lastError = error
}
}
}
throw lastError
case .remote:
let nsError = error as NSError
guard nsError.domain == URLError.errorDomain else { throw error }
var lastError: Error = error
await RemoteTunnelManager.shared.stopAll()
do {
_ = try await GatewayEndpointStore.shared.ensureRemoteControlTunnel()
} catch {
lastError = error
}
for delayMs in [150, 400, 900] {
try await Task.sleep(nanoseconds: UInt64(delayMs) * 1_000_000)
do {
let cfg = try await self.configProvider()
await self.configure(url: cfg.url, token: cfg.token, password: cfg.password)
guard let client = self.client else {
throw NSError(
domain: "Gateway",
code: 0,
userInfo: [NSLocalizedDescriptionKey: "gateway not configured"])
}
return try await client.request(method: method, params: params, timeoutMs: timeoutMs)
} catch {
lastError = error
}
}
throw lastError
case .unconfigured:
throw error
}
}
}
func requestRaw(
method: Method,
params: [String: AnyCodable]? = nil,
timeoutMs: Double? = nil) async throws -> Data
{
try await self.request(method: method.rawValue, params: params, timeoutMs: timeoutMs)
}
func requestRaw(
method: String,
params: [String: AnyCodable]? = nil,
timeoutMs: Double? = nil) async throws -> Data
{
try await self.request(method: method, params: params, timeoutMs: timeoutMs)
}
func requestDecoded<T: Decodable>(
method: Method,
params: [String: AnyCodable]? = nil,
timeoutMs: Double? = nil) async throws -> T
{
let data = try await self.requestRaw(method: method, params: params, timeoutMs: timeoutMs)
do {
return try self.decoder.decode(T.self, from: data)
} catch {
throw GatewayDecodingError(method: method.rawValue, message: error.localizedDescription)
}
}
func requestVoid(
method: Method,
params: [String: AnyCodable]? = nil,
timeoutMs: Double? = nil) async throws
{
_ = try await self.requestRaw(method: method, params: params, timeoutMs: timeoutMs)
}
/// Ensure the underlying socket is configured (and replaced if config changed).
func refresh() async throws {
let cfg = try await self.configProvider()
await self.configure(url: cfg.url, token: cfg.token, password: cfg.password)
}
func authSource() async -> GatewayAuthSource? {
guard let client else { return nil }
return await client.authSource()
}
func shutdown() async {
if let client {
await client.shutdown()
}
self.client = nil
self.configuredURL = nil
self.configuredToken = nil
self.lastSnapshot = nil
}
func canvasHostUrl() async -> String? {
guard let snapshot = self.lastSnapshot else { return nil }
let trimmed = snapshot.canvashosturl?.trimmingCharacters(in: CharacterSet.whitespacesAndNewlines) ?? ""
return trimmed.isEmpty ? nil : trimmed
}
private func sessionDefaultString(_ defaults: [String: MoltbotProtocol.AnyCodable]?, key: String) -> String {
let raw = defaults?[key]?.value as? String
return (raw ?? "").trimmingCharacters(in: CharacterSet.whitespacesAndNewlines)
}
func cachedMainSessionKey() -> String? {
guard let snapshot = self.lastSnapshot else { return nil }
let trimmed = self.sessionDefaultString(snapshot.snapshot.sessiondefaults, key: "mainSessionKey")
return trimmed.isEmpty ? nil : trimmed
}
func cachedGatewayVersion() -> String? {
guard let snapshot = self.lastSnapshot else { return nil }
let raw = snapshot.server["version"]?.value as? String
let trimmed = raw?.trimmingCharacters(in: CharacterSet.whitespacesAndNewlines) ?? ""
return trimmed.isEmpty ? nil : trimmed
}
func snapshotPaths() -> (configPath: String?, stateDir: String?) {
guard let snapshot = self.lastSnapshot else { return (nil, nil) }
let configPath = snapshot.snapshot.configpath?.trimmingCharacters(in: .whitespacesAndNewlines)
let stateDir = snapshot.snapshot.statedir?.trimmingCharacters(in: .whitespacesAndNewlines)
return (
configPath?.isEmpty == false ? configPath : nil,
stateDir?.isEmpty == false ? stateDir : nil)
}
func subscribe(bufferingNewest: Int = 100) -> AsyncStream<GatewayPush> {
let id = UUID()
let snapshot = self.lastSnapshot
let connection = self
return AsyncStream(bufferingPolicy: .bufferingNewest(bufferingNewest)) { continuation in
if let snapshot {
continuation.yield(.snapshot(snapshot))
}
self.subscribers[id] = continuation
continuation.onTermination = { @Sendable _ in
Task { await connection.removeSubscriber(id) }
}
}
}
private func removeSubscriber(_ id: UUID) {
self.subscribers[id] = nil
}
private func broadcast(_ push: GatewayPush) {
if case let .snapshot(snapshot) = push {
self.lastSnapshot = snapshot
if let mainSessionKey = self.cachedMainSessionKey() {
Task { @MainActor in
WorkActivityStore.shared.setMainSessionKey(mainSessionKey)
}
}
}
for (_, continuation) in self.subscribers {
continuation.yield(push)
}
}
private func canonicalizeSessionKey(_ raw: String) -> String {
let trimmed = raw.trimmingCharacters(in: CharacterSet.whitespacesAndNewlines)
guard !trimmed.isEmpty else { return trimmed }
guard let defaults = self.lastSnapshot?.snapshot.sessiondefaults else { return trimmed }
let mainSessionKey = self.sessionDefaultString(defaults, key: "mainSessionKey")
guard !mainSessionKey.isEmpty else { return trimmed }
let mainKey = self.sessionDefaultString(defaults, key: "mainKey")
let defaultAgentId = self.sessionDefaultString(defaults, key: "defaultAgentId")
let isMainAlias =
trimmed == "main" ||
(!mainKey.isEmpty && trimmed == mainKey) ||
trimmed == mainSessionKey ||
(!defaultAgentId.isEmpty &&
(trimmed == "agent:\(defaultAgentId):main" ||
(mainKey.isEmpty == false && trimmed == "agent:\(defaultAgentId):\(mainKey)")))
return isMainAlias ? mainSessionKey : trimmed
}
private func configure(url: URL, token: String?, password: String?) async {
if self.client != nil, self.configuredURL == url, self.configuredToken == token,
self.configuredPassword == password
{
return
}
if let client {
await client.shutdown()
}
self.lastSnapshot = nil
self.client = GatewayChannelActor(
url: url,
token: token,
password: password,
session: self.sessionBox,
pushHandler: { [weak self] push in
await self?.handle(push: push)
})
self.configuredURL = url
self.configuredToken = token
self.configuredPassword = password
}
private func handle(push: GatewayPush) {
self.broadcast(push)
}
private static func defaultConfigProvider() async throws -> Config {
try await GatewayEndpointStore.shared.requireConfig()
}
}
// MARK: - Typed gateway API
extension GatewayConnection {
struct ConfigGetSnapshot: Decodable, Sendable {
struct SnapshotConfig: Decodable, Sendable {
struct Session: Decodable, Sendable {
let mainKey: String?
let scope: String?
}
let session: Session?
}
let config: SnapshotConfig?
}
static func mainSessionKey(fromConfigGetData data: Data) throws -> String {
let snapshot = try JSONDecoder().decode(ConfigGetSnapshot.self, from: data)
let scope = snapshot.config?.session?.scope?.trimmingCharacters(in: .whitespacesAndNewlines)
if scope == "global" {
return "global"
}
return "main"
}
func mainSessionKey(timeoutMs: Double = 15000) async -> String {
if let cached = self.cachedMainSessionKey() {
return cached
}
do {
let data = try await self.requestRaw(method: "config.get", params: nil, timeoutMs: timeoutMs)
return try Self.mainSessionKey(fromConfigGetData: data)
} catch {
return "main"
}
}
func status() async -> (ok: Bool, error: String?) {
do {
_ = try await self.requestRaw(method: .status)
return (true, nil)
} catch {
return (false, error.localizedDescription)
}
}
func setHeartbeatsEnabled(_ enabled: Bool) async -> Bool {
do {
try await self.requestVoid(method: .setHeartbeats, params: ["enabled": AnyCodable(enabled)])
return true
} catch {
gatewayConnectionLogger.error("setHeartbeatsEnabled failed \(error.localizedDescription, privacy: .public)")
return false
}
}
func sendAgent(_ invocation: GatewayAgentInvocation) async -> (ok: Bool, error: String?) {
let trimmed = invocation.message.trimmingCharacters(in: .whitespacesAndNewlines)
guard !trimmed.isEmpty else { return (false, "message empty") }
let sessionKey = self.canonicalizeSessionKey(invocation.sessionKey)
var params: [String: AnyCodable] = [
"message": AnyCodable(trimmed),
"sessionKey": AnyCodable(sessionKey),
"thinking": AnyCodable(invocation.thinking ?? "default"),
"deliver": AnyCodable(invocation.deliver),
"to": AnyCodable(invocation.to ?? ""),
"channel": AnyCodable(invocation.channel.rawValue),
"idempotencyKey": AnyCodable(invocation.idempotencyKey),
]
if let timeout = invocation.timeoutSeconds {
params["timeout"] = AnyCodable(timeout)
}
do {
try await self.requestVoid(method: .agent, params: params)
return (true, nil)
} catch {
return (false, error.localizedDescription)
}
}
func sendAgent(
message: String,
thinking: String?,
sessionKey: String,
deliver: Bool,
to: String?,
channel: GatewayAgentChannel = .last,
timeoutSeconds: Int? = nil,
idempotencyKey: String = UUID().uuidString) async -> (ok: Bool, error: String?)
{
await self.sendAgent(GatewayAgentInvocation(
message: message,
sessionKey: sessionKey,
thinking: thinking,
deliver: deliver,
to: to,
channel: channel,
timeoutSeconds: timeoutSeconds,
idempotencyKey: idempotencyKey))
}
func sendSystemEvent(_ params: [String: AnyCodable]) async {
do {
try await self.requestVoid(method: .systemEvent, params: params)
} catch {
// Best-effort only.
}
}
// MARK: - Health
func healthSnapshot(timeoutMs: Double? = nil) async throws -> HealthSnapshot {
let data = try await self.requestRaw(method: .health, timeoutMs: timeoutMs)
if let snap = decodeHealthSnapshot(from: data) { return snap }
throw GatewayDecodingError(method: Method.health.rawValue, message: "failed to decode health snapshot")
}
func healthOK(timeoutMs: Int = 8000) async throws -> Bool {
let data = try await self.requestRaw(method: .health, timeoutMs: Double(timeoutMs))
return (try? self.decoder.decode(MoltbotGatewayHealthOK.self, from: data))?.ok ?? true
}
// MARK: - Skills
func skillsStatus() async throws -> SkillsStatusReport {
try await self.requestDecoded(method: .skillsStatus)
}
func skillsInstall(
name: String,
installId: String,
timeoutMs: Int? = nil) async throws -> SkillInstallResult
{
var params: [String: AnyCodable] = [
"name": AnyCodable(name),
"installId": AnyCodable(installId),
]
if let timeoutMs {
params["timeoutMs"] = AnyCodable(timeoutMs)
}
return try await self.requestDecoded(method: .skillsInstall, params: params)
}
func skillsUpdate(
skillKey: String,
enabled: Bool? = nil,
apiKey: String? = nil,
env: [String: String]? = nil) async throws -> SkillUpdateResult
{
var params: [String: AnyCodable] = [
"skillKey": AnyCodable(skillKey),
]
if let enabled { params["enabled"] = AnyCodable(enabled) }
if let apiKey { params["apiKey"] = AnyCodable(apiKey) }
if let env, !env.isEmpty { params["env"] = AnyCodable(env) }
return try await self.requestDecoded(method: .skillsUpdate, params: params)
}
// MARK: - Sessions
func sessionsPreview(
keys: [String],
limit: Int? = nil,
maxChars: Int? = nil,
timeoutMs: Int? = nil) async throws -> MoltbotSessionsPreviewPayload
{
let resolvedKeys = keys
.map { self.canonicalizeSessionKey($0) }
.filter { !$0.isEmpty }
if resolvedKeys.isEmpty {
return MoltbotSessionsPreviewPayload(ts: 0, previews: [])
}
var params: [String: AnyCodable] = ["keys": AnyCodable(resolvedKeys)]
if let limit { params["limit"] = AnyCodable(limit) }
if let maxChars { params["maxChars"] = AnyCodable(maxChars) }
let timeout = timeoutMs.map { Double($0) }
return try await self.requestDecoded(
method: .sessionsPreview,
params: params,
timeoutMs: timeout)
}
// MARK: - Chat
func chatHistory(
sessionKey: String,
limit: Int? = nil,
timeoutMs: Int? = nil) async throws -> MoltbotChatHistoryPayload
{
let resolvedKey = self.canonicalizeSessionKey(sessionKey)
var params: [String: AnyCodable] = ["sessionKey": AnyCodable(resolvedKey)]
if let limit { params["limit"] = AnyCodable(limit) }
let timeout = timeoutMs.map { Double($0) }
return try await self.requestDecoded(
method: .chatHistory,
params: params,
timeoutMs: timeout)
}
func chatSend(
sessionKey: String,
message: String,
thinking: String,
idempotencyKey: String,
attachments: [MoltbotChatAttachmentPayload],
timeoutMs: Int = 30000) async throws -> MoltbotChatSendResponse
{
let resolvedKey = self.canonicalizeSessionKey(sessionKey)
var params: [String: AnyCodable] = [
"sessionKey": AnyCodable(resolvedKey),
"message": AnyCodable(message),
"thinking": AnyCodable(thinking),
"idempotencyKey": AnyCodable(idempotencyKey),
"timeoutMs": AnyCodable(timeoutMs),
]
if !attachments.isEmpty {
let encoded = attachments.map { att in
[
"type": att.type,
"mimeType": att.mimeType,
"fileName": att.fileName,
"content": att.content,
]
}
params["attachments"] = AnyCodable(encoded)
}
return try await self.requestDecoded(
method: .chatSend,
params: params,
timeoutMs: Double(timeoutMs))
}
func chatAbort(sessionKey: String, runId: String) async throws -> Bool {
let resolvedKey = self.canonicalizeSessionKey(sessionKey)
struct AbortResponse: Decodable { let ok: Bool?; let aborted: Bool? }
let res: AbortResponse = try await self.requestDecoded(
method: .chatAbort,
params: ["sessionKey": AnyCodable(resolvedKey), "runId": AnyCodable(runId)])
return res.aborted ?? false
}
func talkMode(enabled: Bool, phase: String? = nil) async {
var params: [String: AnyCodable] = ["enabled": AnyCodable(enabled)]
if let phase { params["phase"] = AnyCodable(phase) }
try? await self.requestVoid(method: .talkMode, params: params)
}
// MARK: - VoiceWake
func voiceWakeGetTriggers() async throws -> [String] {
struct VoiceWakePayload: Decodable { let triggers: [String] }
let payload: VoiceWakePayload = try await self.requestDecoded(method: .voicewakeGet)
return payload.triggers
}
func voiceWakeSetTriggers(_ triggers: [String]) async {
do {
try await self.requestVoid(
method: .voicewakeSet,
params: ["triggers": AnyCodable(triggers)],
timeoutMs: 10000)
} catch {
// Best-effort only.
}
}
// MARK: - Node pairing
func nodePairApprove(requestId: String) async throws {
try await self.requestVoid(
method: .nodePairApprove,
params: ["requestId": AnyCodable(requestId)],
timeoutMs: 10000)
}
func nodePairReject(requestId: String) async throws {
try await self.requestVoid(
method: .nodePairReject,
params: ["requestId": AnyCodable(requestId)],
timeoutMs: 10000)
}
// MARK: - Device pairing
func devicePairApprove(requestId: String) async throws {
try await self.requestVoid(
method: .devicePairApprove,
params: ["requestId": AnyCodable(requestId)],
timeoutMs: 10000)
}
func devicePairReject(requestId: String) async throws {
try await self.requestVoid(
method: .devicePairReject,
params: ["requestId": AnyCodable(requestId)],
timeoutMs: 10000)
}
// MARK: - Cron
struct CronSchedulerStatus: Decodable, Sendable {
let enabled: Bool
let storePath: String
let jobs: Int
let nextWakeAtMs: Int?
}
func cronStatus() async throws -> CronSchedulerStatus {
try await self.requestDecoded(method: .cronStatus)
}
func cronList(includeDisabled: Bool = true) async throws -> [CronJob] {
let res: CronListResponse = try await self.requestDecoded(
method: .cronList,
params: ["includeDisabled": AnyCodable(includeDisabled)])
return res.jobs
}
func cronRuns(jobId: String, limit: Int = 200) async throws -> [CronRunLogEntry] {
let res: CronRunsResponse = try await self.requestDecoded(
method: .cronRuns,
params: ["id": AnyCodable(jobId), "limit": AnyCodable(limit)])
return res.entries
}
func cronRun(jobId: String, force: Bool = true) async throws {
try await self.requestVoid(
method: .cronRun,
params: [
"id": AnyCodable(jobId),
"mode": AnyCodable(force ? "force" : "due"),
],
timeoutMs: 20000)
}
func cronRemove(jobId: String) async throws {
try await self.requestVoid(method: .cronRemove, params: ["id": AnyCodable(jobId)])
}
func cronUpdate(jobId: String, patch: [String: AnyCodable]) async throws {
try await self.requestVoid(
method: .cronUpdate,
params: ["id": AnyCodable(jobId), "patch": AnyCodable(patch)])
}
func cronAdd(payload: [String: AnyCodable]) async throws {
try await self.requestVoid(method: .cronAdd, params: payload)
}
}

View File

@@ -0,0 +1,63 @@
import Foundation
import Observation
import OSLog
@MainActor
@Observable
final class GatewayConnectivityCoordinator {
static let shared = GatewayConnectivityCoordinator()
private let logger = Logger(subsystem: "bot.molt", category: "gateway.connectivity")
private var endpointTask: Task<Void, Never>?
private var lastResolvedURL: URL?
private(set) var endpointState: GatewayEndpointState?
private(set) var resolvedURL: URL?
private(set) var resolvedMode: AppState.ConnectionMode?
private(set) var resolvedHostLabel: String?
private init() {
self.start()
}
func start() {
guard self.endpointTask == nil else { return }
self.endpointTask = Task { [weak self] in
guard let self else { return }
let stream = await GatewayEndpointStore.shared.subscribe()
for await state in stream {
await MainActor.run { self.handleEndpointState(state) }
}
}
}
var localEndpointHostLabel: String? {
guard self.resolvedMode == .local, let url = self.resolvedURL else { return nil }
return Self.hostLabel(for: url)
}
private func handleEndpointState(_ state: GatewayEndpointState) {
self.endpointState = state
switch state {
case let .ready(mode, url, _, _):
self.resolvedMode = mode
self.resolvedURL = url
self.resolvedHostLabel = Self.hostLabel(for: url)
let urlChanged = self.lastResolvedURL?.absoluteString != url.absoluteString
if urlChanged {
self.lastResolvedURL = url
Task { await ControlChannel.shared.refreshEndpoint(reason: "endpoint changed") }
}
case let .connecting(mode, _):
self.resolvedMode = mode
case let .unavailable(mode, _):
self.resolvedMode = mode
}
}
private static func hostLabel(for url: URL) -> String {
let host = url.host ?? url.absoluteString
if let port = url.port { return "\(host):\(port)" }
return host
}
}

View File

@@ -0,0 +1,696 @@
import ConcurrencyExtras
import Foundation
import OSLog
enum GatewayEndpointState: Sendable, Equatable {
case ready(mode: AppState.ConnectionMode, url: URL, token: String?, password: String?)
case connecting(mode: AppState.ConnectionMode, detail: String)
case unavailable(mode: AppState.ConnectionMode, reason: String)
}
/// Single place to resolve (and publish) the effective gateway control endpoint.
///
/// This is intentionally separate from `GatewayConnection`:
/// - `GatewayConnection` consumes the resolved endpoint (no tunnel side-effects).
/// - The endpoint store owns observation + explicit "ensure tunnel" actions.
actor GatewayEndpointStore {
static let shared = GatewayEndpointStore()
private static let supportedBindModes: Set<String> = [
"loopback",
"tailnet",
"lan",
"auto",
"custom",
]
private static let remoteConnectingDetail = "Connecting to remote gateway…"
private static let staticLogger = Logger(subsystem: "bot.molt", category: "gateway-endpoint")
private enum EnvOverrideWarningKind: Sendable {
case token
case password
}
private static let envOverrideWarnings = LockIsolated((token: false, password: false))
struct Deps: Sendable {
let mode: @Sendable () async -> AppState.ConnectionMode
let token: @Sendable () -> String?
let password: @Sendable () -> String?
let localPort: @Sendable () -> Int
let localHost: @Sendable () async -> String
let remotePortIfRunning: @Sendable () async -> UInt16?
let ensureRemoteTunnel: @Sendable () async throws -> UInt16
static let live = Deps(
mode: { await MainActor.run { AppStateStore.shared.connectionMode } },
token: {
let root = MoltbotConfigFile.loadDict()
let isRemote = ConnectionModeResolver.resolve(root: root).mode == .remote
return GatewayEndpointStore.resolveGatewayToken(
isRemote: isRemote,
root: root,
env: ProcessInfo.processInfo.environment,
launchdSnapshot: GatewayLaunchAgentManager.launchdConfigSnapshot())
},
password: {
let root = MoltbotConfigFile.loadDict()
let isRemote = ConnectionModeResolver.resolve(root: root).mode == .remote
return GatewayEndpointStore.resolveGatewayPassword(
isRemote: isRemote,
root: root,
env: ProcessInfo.processInfo.environment,
launchdSnapshot: GatewayLaunchAgentManager.launchdConfigSnapshot())
},
localPort: { GatewayEnvironment.gatewayPort() },
localHost: {
let root = MoltbotConfigFile.loadDict()
let bind = GatewayEndpointStore.resolveGatewayBindMode(
root: root,
env: ProcessInfo.processInfo.environment)
let customBindHost = GatewayEndpointStore.resolveGatewayCustomBindHost(root: root)
let tailscaleIP = await MainActor.run { TailscaleService.shared.tailscaleIP }
?? TailscaleService.fallbackTailnetIPv4()
return GatewayEndpointStore.resolveLocalGatewayHost(
bindMode: bind,
customBindHost: customBindHost,
tailscaleIP: tailscaleIP)
},
remotePortIfRunning: { await RemoteTunnelManager.shared.controlTunnelPortIfRunning() },
ensureRemoteTunnel: { try await RemoteTunnelManager.shared.ensureControlTunnel() })
}
private static func resolveGatewayPassword(
isRemote: Bool,
root: [String: Any],
env: [String: String],
launchdSnapshot: LaunchAgentPlistSnapshot?) -> String?
{
let raw = env["CLAWDBOT_GATEWAY_PASSWORD"] ?? ""
let trimmed = raw.trimmingCharacters(in: .whitespacesAndNewlines)
if !trimmed.isEmpty {
if let configPassword = self.resolveConfigPassword(isRemote: isRemote, root: root),
!configPassword.isEmpty
{
self.warnEnvOverrideOnce(
kind: .password,
envVar: "CLAWDBOT_GATEWAY_PASSWORD",
configKey: isRemote ? "gateway.remote.password" : "gateway.auth.password")
}
return trimmed
}
if isRemote {
if let gateway = root["gateway"] as? [String: Any],
let remote = gateway["remote"] as? [String: Any],
let password = remote["password"] as? String
{
let pw = password.trimmingCharacters(in: .whitespacesAndNewlines)
if !pw.isEmpty {
return pw
}
}
return nil
}
if let gateway = root["gateway"] as? [String: Any],
let auth = gateway["auth"] as? [String: Any],
let password = auth["password"] as? String
{
let pw = password.trimmingCharacters(in: .whitespacesAndNewlines)
if !pw.isEmpty {
return pw
}
}
if let password = launchdSnapshot?.password?.trimmingCharacters(in: .whitespacesAndNewlines),
!password.isEmpty
{
return password
}
return nil
}
private static func resolveConfigPassword(isRemote: Bool, root: [String: Any]) -> String? {
if isRemote {
if let gateway = root["gateway"] as? [String: Any],
let remote = gateway["remote"] as? [String: Any],
let password = remote["password"] as? String
{
return password.trimmingCharacters(in: .whitespacesAndNewlines)
}
return nil
}
if let gateway = root["gateway"] as? [String: Any],
let auth = gateway["auth"] as? [String: Any],
let password = auth["password"] as? String
{
return password.trimmingCharacters(in: .whitespacesAndNewlines)
}
return nil
}
private static func resolveGatewayToken(
isRemote: Bool,
root: [String: Any],
env: [String: String],
launchdSnapshot: LaunchAgentPlistSnapshot?) -> String?
{
let raw = env["CLAWDBOT_GATEWAY_TOKEN"] ?? ""
let trimmed = raw.trimmingCharacters(in: .whitespacesAndNewlines)
if !trimmed.isEmpty {
if let configToken = self.resolveConfigToken(isRemote: isRemote, root: root),
!configToken.isEmpty,
configToken != trimmed
{
self.warnEnvOverrideOnce(
kind: .token,
envVar: "CLAWDBOT_GATEWAY_TOKEN",
configKey: isRemote ? "gateway.remote.token" : "gateway.auth.token")
}
return trimmed
}
if let configToken = self.resolveConfigToken(isRemote: isRemote, root: root),
!configToken.isEmpty
{
return configToken
}
if isRemote {
return nil
}
if let token = launchdSnapshot?.token?.trimmingCharacters(in: .whitespacesAndNewlines),
!token.isEmpty
{
return token
}
return nil
}
private static func resolveConfigToken(isRemote: Bool, root: [String: Any]) -> String? {
if isRemote {
if let gateway = root["gateway"] as? [String: Any],
let remote = gateway["remote"] as? [String: Any],
let token = remote["token"] as? String
{
return token.trimmingCharacters(in: .whitespacesAndNewlines)
}
return nil
}
if let gateway = root["gateway"] as? [String: Any],
let auth = gateway["auth"] as? [String: Any],
let token = auth["token"] as? String
{
return token.trimmingCharacters(in: .whitespacesAndNewlines)
}
return nil
}
private static func warnEnvOverrideOnce(
kind: EnvOverrideWarningKind,
envVar: String,
configKey: String)
{
let shouldWarn = Self.envOverrideWarnings.withValue { state in
switch kind {
case .token:
guard !state.token else { return false }
state.token = true
return true
case .password:
guard !state.password else { return false }
state.password = true
return true
}
}
guard shouldWarn else { return }
Self.staticLogger.warning(
"\(envVar, privacy: .public) is set and overrides \(configKey, privacy: .public). " +
"If this is unintentional, clear it with: launchctl unsetenv \(envVar, privacy: .public)")
}
private let deps: Deps
private let logger = Logger(subsystem: "bot.molt", category: "gateway-endpoint")
private var state: GatewayEndpointState
private var subscribers: [UUID: AsyncStream<GatewayEndpointState>.Continuation] = [:]
private var remoteEnsure: (token: UUID, task: Task<UInt16, Error>)?
init(deps: Deps = .live) {
self.deps = deps
let modeRaw = UserDefaults.standard.string(forKey: connectionModeKey)
let initialMode: AppState.ConnectionMode
if let modeRaw {
initialMode = AppState.ConnectionMode(rawValue: modeRaw) ?? .local
} else {
let seen = UserDefaults.standard.bool(forKey: "moltbot.onboardingSeen")
initialMode = seen ? .local : .unconfigured
}
let port = deps.localPort()
let bind = GatewayEndpointStore.resolveGatewayBindMode(
root: MoltbotConfigFile.loadDict(),
env: ProcessInfo.processInfo.environment)
let customBindHost = GatewayEndpointStore.resolveGatewayCustomBindHost(root: MoltbotConfigFile.loadDict())
let scheme = GatewayEndpointStore.resolveGatewayScheme(
root: MoltbotConfigFile.loadDict(),
env: ProcessInfo.processInfo.environment)
let host = GatewayEndpointStore.resolveLocalGatewayHost(
bindMode: bind,
customBindHost: customBindHost,
tailscaleIP: nil)
let token = deps.token()
let password = deps.password()
switch initialMode {
case .local:
self.state = .ready(
mode: .local,
url: URL(string: "\(scheme)://\(host):\(port)")!,
token: token,
password: password)
case .remote:
self.state = .connecting(mode: .remote, detail: Self.remoteConnectingDetail)
Task { await self.setMode(.remote) }
case .unconfigured:
self.state = .unavailable(mode: .unconfigured, reason: "Gateway not configured")
}
}
func subscribe(bufferingNewest: Int = 1) -> AsyncStream<GatewayEndpointState> {
let id = UUID()
let initial = self.state
let store = self
return AsyncStream(bufferingPolicy: .bufferingNewest(bufferingNewest)) { continuation in
continuation.yield(initial)
self.subscribers[id] = continuation
continuation.onTermination = { @Sendable _ in
Task { await store.removeSubscriber(id) }
}
}
}
func refresh() async {
let mode = await self.deps.mode()
await self.setMode(mode)
}
func setMode(_ mode: AppState.ConnectionMode) async {
let token = self.deps.token()
let password = self.deps.password()
switch mode {
case .local:
self.cancelRemoteEnsure()
let port = self.deps.localPort()
let host = await self.deps.localHost()
let scheme = GatewayEndpointStore.resolveGatewayScheme(
root: MoltbotConfigFile.loadDict(),
env: ProcessInfo.processInfo.environment)
self.setState(.ready(
mode: .local,
url: URL(string: "\(scheme)://\(host):\(port)")!,
token: token,
password: password))
case .remote:
let root = MoltbotConfigFile.loadDict()
if GatewayRemoteConfig.resolveTransport(root: root) == .direct {
guard let url = GatewayRemoteConfig.resolveGatewayUrl(root: root) else {
self.cancelRemoteEnsure()
self.setState(.unavailable(
mode: .remote,
reason: "gateway.remote.url missing or invalid for direct transport"))
return
}
self.cancelRemoteEnsure()
self.setState(.ready(mode: .remote, url: url, token: token, password: password))
return
}
let port = await self.deps.remotePortIfRunning()
guard let port else {
self.setState(.connecting(mode: .remote, detail: Self.remoteConnectingDetail))
self.kickRemoteEnsureIfNeeded(detail: Self.remoteConnectingDetail)
return
}
self.cancelRemoteEnsure()
let scheme = GatewayEndpointStore.resolveGatewayScheme(
root: MoltbotConfigFile.loadDict(),
env: ProcessInfo.processInfo.environment)
self.setState(.ready(
mode: .remote,
url: URL(string: "\(scheme)://127.0.0.1:\(Int(port))")!,
token: token,
password: password))
case .unconfigured:
self.cancelRemoteEnsure()
self.setState(.unavailable(mode: .unconfigured, reason: "Gateway not configured"))
}
}
/// Explicit action: ensure the remote control tunnel is established and publish the resolved endpoint.
func ensureRemoteControlTunnel() async throws -> UInt16 {
let mode = await self.deps.mode()
guard mode == .remote else {
throw NSError(
domain: "RemoteTunnel",
code: 1,
userInfo: [NSLocalizedDescriptionKey: "Remote mode is not enabled"])
}
let root = MoltbotConfigFile.loadDict()
if GatewayRemoteConfig.resolveTransport(root: root) == .direct {
guard let url = GatewayRemoteConfig.resolveGatewayUrl(root: root) else {
throw NSError(
domain: "GatewayEndpoint",
code: 1,
userInfo: [NSLocalizedDescriptionKey: "gateway.remote.url missing or invalid"])
}
guard let port = GatewayRemoteConfig.defaultPort(for: url),
let portInt = UInt16(exactly: port)
else {
throw NSError(
domain: "GatewayEndpoint",
code: 1,
userInfo: [NSLocalizedDescriptionKey: "Invalid gateway.remote.url port"])
}
self.logger.info("remote transport direct; skipping SSH tunnel")
return portInt
}
let config = try await self.ensureRemoteConfig(detail: Self.remoteConnectingDetail)
guard let portInt = config.0.port, let port = UInt16(exactly: portInt) else {
throw NSError(
domain: "GatewayEndpoint",
code: 1,
userInfo: [NSLocalizedDescriptionKey: "Missing tunnel port"])
}
return port
}
func requireConfig() async throws -> GatewayConnection.Config {
await self.refresh()
switch self.state {
case let .ready(_, url, token, password):
return (url, token, password)
case let .connecting(mode, _):
guard mode == .remote else {
throw NSError(domain: "GatewayEndpoint", code: 1, userInfo: [NSLocalizedDescriptionKey: "Connecting…"])
}
return try await self.ensureRemoteConfig(detail: Self.remoteConnectingDetail)
case let .unavailable(mode, reason):
guard mode == .remote else {
throw NSError(domain: "GatewayEndpoint", code: 1, userInfo: [NSLocalizedDescriptionKey: reason])
}
// Auto-recover for remote mode: if the SSH control tunnel died (or hasn't been created yet),
// recreate it on demand so callers can recover without a manual reconnect.
self.logger.info(
"endpoint unavailable; ensuring remote control tunnel reason=\(reason, privacy: .public)")
return try await self.ensureRemoteConfig(detail: Self.remoteConnectingDetail)
}
}
private func cancelRemoteEnsure() {
self.remoteEnsure?.task.cancel()
self.remoteEnsure = nil
}
private func kickRemoteEnsureIfNeeded(detail: String) {
if self.remoteEnsure != nil {
self.setState(.connecting(mode: .remote, detail: detail))
return
}
let deps = self.deps
let token = UUID()
let task = Task.detached(priority: .utility) { try await deps.ensureRemoteTunnel() }
self.remoteEnsure = (token: token, task: task)
self.setState(.connecting(mode: .remote, detail: detail))
}
private func ensureRemoteConfig(detail: String) async throws -> GatewayConnection.Config {
let mode = await self.deps.mode()
guard mode == .remote else {
throw NSError(
domain: "RemoteTunnel",
code: 1,
userInfo: [NSLocalizedDescriptionKey: "Remote mode is not enabled"])
}
let root = MoltbotConfigFile.loadDict()
if GatewayRemoteConfig.resolveTransport(root: root) == .direct {
guard let url = GatewayRemoteConfig.resolveGatewayUrl(root: root) else {
throw NSError(
domain: "GatewayEndpoint",
code: 1,
userInfo: [NSLocalizedDescriptionKey: "gateway.remote.url missing or invalid"])
}
let token = self.deps.token()
let password = self.deps.password()
self.cancelRemoteEnsure()
self.setState(.ready(mode: .remote, url: url, token: token, password: password))
return (url, token, password)
}
self.kickRemoteEnsureIfNeeded(detail: detail)
guard let ensure = self.remoteEnsure else {
throw NSError(domain: "GatewayEndpoint", code: 1, userInfo: [NSLocalizedDescriptionKey: "Connecting…"])
}
do {
let forwarded = try await ensure.task.value
let stillRemote = await self.deps.mode() == .remote
guard stillRemote else {
throw NSError(
domain: "RemoteTunnel",
code: 1,
userInfo: [NSLocalizedDescriptionKey: "Remote mode is not enabled"])
}
if self.remoteEnsure?.token == ensure.token {
self.remoteEnsure = nil
}
let token = self.deps.token()
let password = self.deps.password()
let scheme = GatewayEndpointStore.resolveGatewayScheme(
root: MoltbotConfigFile.loadDict(),
env: ProcessInfo.processInfo.environment)
let url = URL(string: "\(scheme)://127.0.0.1:\(Int(forwarded))")!
self.setState(.ready(mode: .remote, url: url, token: token, password: password))
return (url, token, password)
} catch let err as CancellationError {
if self.remoteEnsure?.token == ensure.token {
self.remoteEnsure = nil
}
throw err
} catch {
if self.remoteEnsure?.token == ensure.token {
self.remoteEnsure = nil
}
let msg = "Remote control tunnel failed (\(error.localizedDescription))"
self.setState(.unavailable(mode: .remote, reason: msg))
self.logger.error("remote control tunnel ensure failed \(msg, privacy: .public)")
throw NSError(domain: "GatewayEndpoint", code: 1, userInfo: [NSLocalizedDescriptionKey: msg])
}
}
private func removeSubscriber(_ id: UUID) {
self.subscribers[id] = nil
}
private func setState(_ next: GatewayEndpointState) {
guard next != self.state else { return }
self.state = next
for (_, continuation) in self.subscribers {
continuation.yield(next)
}
switch next {
case let .ready(mode, url, _, _):
let modeDesc = String(describing: mode)
let urlDesc = url.absoluteString
self.logger
.debug(
"resolved endpoint mode=\(modeDesc, privacy: .public) url=\(urlDesc, privacy: .public)")
case let .connecting(mode, detail):
let modeDesc = String(describing: mode)
self.logger
.debug(
"endpoint connecting mode=\(modeDesc, privacy: .public) detail=\(detail, privacy: .public)")
case let .unavailable(mode, reason):
let modeDesc = String(describing: mode)
self.logger
.debug(
"endpoint unavailable mode=\(modeDesc, privacy: .public) reason=\(reason, privacy: .public)")
}
}
func maybeFallbackToTailnet(from currentURL: URL) async -> GatewayConnection.Config? {
let mode = await self.deps.mode()
guard mode == .local else { return nil }
let root = MoltbotConfigFile.loadDict()
let bind = GatewayEndpointStore.resolveGatewayBindMode(
root: root,
env: ProcessInfo.processInfo.environment)
guard bind == "tailnet" else { return nil }
let currentHost = currentURL.host?.lowercased() ?? ""
guard currentHost == "127.0.0.1" || currentHost == "localhost" else { return nil }
let tailscaleIP = await MainActor.run { TailscaleService.shared.tailscaleIP }
?? TailscaleService.fallbackTailnetIPv4()
guard let tailscaleIP, !tailscaleIP.isEmpty else { return nil }
let scheme = GatewayEndpointStore.resolveGatewayScheme(
root: root,
env: ProcessInfo.processInfo.environment)
let port = self.deps.localPort()
let token = self.deps.token()
let password = self.deps.password()
let url = URL(string: "\(scheme)://\(tailscaleIP):\(port)")!
self.logger.info("auto bind fallback to tailnet host=\(tailscaleIP, privacy: .public)")
self.setState(.ready(mode: .local, url: url, token: token, password: password))
return (url, token, password)
}
private static func resolveGatewayBindMode(
root: [String: Any],
env: [String: String]) -> String?
{
if let envBind = env["CLAWDBOT_GATEWAY_BIND"] {
let trimmed = envBind.trimmingCharacters(in: .whitespacesAndNewlines).lowercased()
if self.supportedBindModes.contains(trimmed) {
return trimmed
}
}
if let gateway = root["gateway"] as? [String: Any],
let bind = gateway["bind"] as? String
{
let trimmed = bind.trimmingCharacters(in: .whitespacesAndNewlines).lowercased()
if self.supportedBindModes.contains(trimmed) {
return trimmed
}
}
return nil
}
private static func resolveGatewayCustomBindHost(root: [String: Any]) -> String? {
if let gateway = root["gateway"] as? [String: Any],
let customBindHost = gateway["customBindHost"] as? String
{
let trimmed = customBindHost.trimmingCharacters(in: .whitespacesAndNewlines)
return trimmed.isEmpty ? nil : trimmed
}
return nil
}
private static func resolveGatewayScheme(
root: [String: Any],
env: [String: String]) -> String
{
if let envValue = env["CLAWDBOT_GATEWAY_TLS"]?.trimmingCharacters(in: .whitespacesAndNewlines),
!envValue.isEmpty
{
return (envValue == "1" || envValue.lowercased() == "true") ? "wss" : "ws"
}
if let gateway = root["gateway"] as? [String: Any],
let tls = gateway["tls"] as? [String: Any],
let enabled = tls["enabled"] as? Bool
{
return enabled ? "wss" : "ws"
}
return "ws"
}
private static func resolveLocalGatewayHost(
bindMode: String?,
customBindHost: String?,
tailscaleIP: String?) -> String
{
switch bindMode {
case "tailnet":
tailscaleIP ?? "127.0.0.1"
case "auto":
"127.0.0.1"
case "custom":
customBindHost ?? "127.0.0.1"
default:
"127.0.0.1"
}
}
}
extension GatewayEndpointStore {
static func dashboardURL(for config: GatewayConnection.Config) throws -> URL {
guard var components = URLComponents(url: config.url, resolvingAgainstBaseURL: false) else {
throw NSError(domain: "Dashboard", code: 1, userInfo: [
NSLocalizedDescriptionKey: "Invalid gateway URL",
])
}
switch components.scheme?.lowercased() {
case "ws":
components.scheme = "http"
case "wss":
components.scheme = "https"
default:
components.scheme = "http"
}
components.path = "/"
var queryItems: [URLQueryItem] = []
if let token = config.token?.trimmingCharacters(in: .whitespacesAndNewlines),
!token.isEmpty
{
queryItems.append(URLQueryItem(name: "token", value: token))
}
if let password = config.password?.trimmingCharacters(in: .whitespacesAndNewlines),
!password.isEmpty
{
queryItems.append(URLQueryItem(name: "password", value: password))
}
components.queryItems = queryItems.isEmpty ? nil : queryItems
guard let url = components.url else {
throw NSError(domain: "Dashboard", code: 2, userInfo: [
NSLocalizedDescriptionKey: "Failed to build dashboard URL",
])
}
return url
}
}
#if DEBUG
extension GatewayEndpointStore {
static func _testResolveGatewayPassword(
isRemote: Bool,
root: [String: Any],
env: [String: String],
launchdSnapshot: LaunchAgentPlistSnapshot? = nil) -> String?
{
self.resolveGatewayPassword(isRemote: isRemote, root: root, env: env, launchdSnapshot: launchdSnapshot)
}
static func _testResolveGatewayToken(
isRemote: Bool,
root: [String: Any],
env: [String: String],
launchdSnapshot: LaunchAgentPlistSnapshot? = nil) -> String?
{
self.resolveGatewayToken(isRemote: isRemote, root: root, env: env, launchdSnapshot: launchdSnapshot)
}
static func _testResolveGatewayBindMode(
root: [String: Any],
env: [String: String]) -> String?
{
self.resolveGatewayBindMode(root: root, env: env)
}
static func _testResolveLocalGatewayHost(
bindMode: String?,
tailscaleIP: String?,
customBindHost: String? = nil) -> String
{
self.resolveLocalGatewayHost(
bindMode: bindMode,
customBindHost: customBindHost,
tailscaleIP: tailscaleIP)
}
}
#endif

View File

@@ -0,0 +1,342 @@
import MoltbotIPC
import Foundation
import OSLog
// Lightweight SemVer helper (major.minor.patch only) for gateway compatibility checks.
struct Semver: Comparable, CustomStringConvertible, Sendable {
let major: Int
let minor: Int
let patch: Int
var description: String { "\(self.major).\(self.minor).\(self.patch)" }
static func < (lhs: Semver, rhs: Semver) -> Bool {
if lhs.major != rhs.major { return lhs.major < rhs.major }
if lhs.minor != rhs.minor { return lhs.minor < rhs.minor }
return lhs.patch < rhs.patch
}
static func parse(_ raw: String?) -> Semver? {
guard let raw, !raw.isEmpty else { return nil }
let cleaned = raw.trimmingCharacters(in: .whitespacesAndNewlines)
.replacingOccurrences(of: "^v", with: "", options: .regularExpression)
let parts = cleaned.split(separator: ".")
guard parts.count >= 3,
let major = Int(parts[0]),
let minor = Int(parts[1])
else { return nil }
// Strip prerelease suffix (e.g., "11-4" "11", "5-beta.1" "5")
let patchRaw = String(parts[2])
guard let patchToken = patchRaw.split(whereSeparator: { $0 == "-" || $0 == "+" }).first,
let patchNumeric = Int(patchToken)
else {
return nil
}
return Semver(major: major, minor: minor, patch: patchNumeric)
}
func compatible(with required: Semver) -> Bool {
// Same major and not older than required.
self.major == required.major && self >= required
}
}
enum GatewayEnvironmentKind: Equatable {
case checking
case ok
case missingNode
case missingGateway
case incompatible(found: String, required: String)
case error(String)
}
struct GatewayEnvironmentStatus: Equatable {
let kind: GatewayEnvironmentKind
let nodeVersion: String?
let gatewayVersion: String?
let requiredGateway: String?
let message: String
static var checking: Self {
.init(kind: .checking, nodeVersion: nil, gatewayVersion: nil, requiredGateway: nil, message: "Checking…")
}
}
struct GatewayCommandResolution {
let status: GatewayEnvironmentStatus
let command: [String]?
}
enum GatewayEnvironment {
private static let logger = Logger(subsystem: "bot.molt", category: "gateway.env")
private static let supportedBindModes: Set<String> = ["loopback", "tailnet", "lan", "auto"]
static func gatewayPort() -> Int {
if let raw = ProcessInfo.processInfo.environment["CLAWDBOT_GATEWAY_PORT"] {
let trimmed = raw.trimmingCharacters(in: .whitespacesAndNewlines)
if let parsed = Int(trimmed), parsed > 0 { return parsed }
}
if let configPort = MoltbotConfigFile.gatewayPort(), configPort > 0 {
return configPort
}
let stored = UserDefaults.standard.integer(forKey: "gatewayPort")
return stored > 0 ? stored : 18789
}
static func expectedGatewayVersion() -> Semver? {
Semver.parse(self.expectedGatewayVersionString())
}
static func expectedGatewayVersionString() -> String? {
let bundleVersion = Bundle.main.infoDictionary?["CFBundleShortVersionString"] as? String
let trimmed = bundleVersion?.trimmingCharacters(in: .whitespacesAndNewlines)
return (trimmed?.isEmpty == false) ? trimmed : nil
}
// Exposed for tests so we can inject fake version checks without rewriting bundle metadata.
static func expectedGatewayVersion(from versionString: String?) -> Semver? {
Semver.parse(versionString)
}
static func check() -> GatewayEnvironmentStatus {
let start = Date()
defer {
let elapsedMs = Int(Date().timeIntervalSince(start) * 1000)
if elapsedMs > 500 {
self.logger.warning("gateway env check slow (\(elapsedMs, privacy: .public)ms)")
} else {
self.logger.debug("gateway env check ok (\(elapsedMs, privacy: .public)ms)")
}
}
let expected = self.expectedGatewayVersion()
let expectedString = self.expectedGatewayVersionString()
let projectRoot = CommandResolver.projectRoot()
let projectEntrypoint = CommandResolver.gatewayEntrypoint(in: projectRoot)
switch RuntimeLocator.resolve(searchPaths: CommandResolver.preferredPaths()) {
case let .failure(err):
return GatewayEnvironmentStatus(
kind: .missingNode,
nodeVersion: nil,
gatewayVersion: nil,
requiredGateway: expectedString,
message: RuntimeLocator.describeFailure(err))
case let .success(runtime):
let gatewayBin = CommandResolver.clawdbotExecutable()
if gatewayBin == nil, projectEntrypoint == nil {
return GatewayEnvironmentStatus(
kind: .missingGateway,
nodeVersion: runtime.version.description,
gatewayVersion: nil,
requiredGateway: expectedString,
message: "moltbot CLI not found in PATH; install the CLI.")
}
let installed = gatewayBin.flatMap { self.readGatewayVersion(binary: $0) }
?? self.readLocalGatewayVersion(projectRoot: projectRoot)
if let expected, let installed, !installed.compatible(with: expected) {
let expectedText = expectedString ?? expected.description
return GatewayEnvironmentStatus(
kind: .incompatible(found: installed.description, required: expectedText),
nodeVersion: runtime.version.description,
gatewayVersion: installed.description,
requiredGateway: expectedText,
message: """
Gateway version \(installed.description) is incompatible with app \(expectedText);
install or update the global package.
""")
}
let gatewayLabel = gatewayBin != nil ? "global" : "local"
let gatewayVersionText = installed?.description ?? "unknown"
// Avoid repeating "(local)" twice; if using the local entrypoint, show the path once.
let localPathHint = gatewayBin == nil && projectEntrypoint != nil
? " (local: \(projectEntrypoint ?? "unknown"))"
: ""
let gatewayLabelText = gatewayBin != nil
? "(\(gatewayLabel))"
: localPathHint.isEmpty ? "(\(gatewayLabel))" : localPathHint
return GatewayEnvironmentStatus(
kind: .ok,
nodeVersion: runtime.version.description,
gatewayVersion: gatewayVersionText,
requiredGateway: expectedString,
message: "Node \(runtime.version.description); gateway \(gatewayVersionText) \(gatewayLabelText)")
}
}
static func resolveGatewayCommand() -> GatewayCommandResolution {
let start = Date()
defer {
let elapsedMs = Int(Date().timeIntervalSince(start) * 1000)
if elapsedMs > 500 {
self.logger.warning("gateway command resolve slow (\(elapsedMs, privacy: .public)ms)")
} else {
self.logger.debug("gateway command resolve ok (\(elapsedMs, privacy: .public)ms)")
}
}
let projectRoot = CommandResolver.projectRoot()
let projectEntrypoint = CommandResolver.gatewayEntrypoint(in: projectRoot)
let status = self.check()
let gatewayBin = CommandResolver.clawdbotExecutable()
let runtime = RuntimeLocator.resolve(searchPaths: CommandResolver.preferredPaths())
guard case .ok = status.kind else {
return GatewayCommandResolution(status: status, command: nil)
}
let port = self.gatewayPort()
if let gatewayBin {
let bind = self.preferredGatewayBind() ?? "loopback"
let cmd = [gatewayBin, "gateway-daemon", "--port", "\(port)", "--bind", bind]
return GatewayCommandResolution(status: status, command: cmd)
}
if let entry = projectEntrypoint,
case let .success(resolvedRuntime) = runtime
{
let bind = self.preferredGatewayBind() ?? "loopback"
let cmd = [resolvedRuntime.path, entry, "gateway-daemon", "--port", "\(port)", "--bind", bind]
return GatewayCommandResolution(status: status, command: cmd)
}
return GatewayCommandResolution(status: status, command: nil)
}
private static func preferredGatewayBind() -> String? {
if CommandResolver.connectionModeIsRemote() {
return nil
}
if let env = ProcessInfo.processInfo.environment["CLAWDBOT_GATEWAY_BIND"] {
let trimmed = env.trimmingCharacters(in: .whitespacesAndNewlines).lowercased()
if self.supportedBindModes.contains(trimmed) {
return trimmed
}
}
let root = MoltbotConfigFile.loadDict()
if let gateway = root["gateway"] as? [String: Any],
let bind = gateway["bind"] as? String
{
let trimmed = bind.trimmingCharacters(in: .whitespacesAndNewlines).lowercased()
if self.supportedBindModes.contains(trimmed) {
return trimmed
}
}
return nil
}
static func installGlobal(version: Semver?, statusHandler: @escaping @Sendable (String) -> Void) async {
await self.installGlobal(versionString: version?.description, statusHandler: statusHandler)
}
static func installGlobal(versionString: String?, statusHandler: @escaping @Sendable (String) -> Void) async {
let preferred = CommandResolver.preferredPaths().joined(separator: ":")
let trimmed = versionString?.trimmingCharacters(in: .whitespacesAndNewlines)
let target: String = if let trimmed, !trimmed.isEmpty {
trimmed
} else {
"latest"
}
let npm = CommandResolver.findExecutable(named: "npm")
let pnpm = CommandResolver.findExecutable(named: "pnpm")
let bun = CommandResolver.findExecutable(named: "bun")
let (label, cmd): (String, [String]) =
if let npm {
("npm", [npm, "install", "-g", "moltbot@\(target)"])
} else if let pnpm {
("pnpm", [pnpm, "add", "-g", "moltbot@\(target)"])
} else if let bun {
("bun", [bun, "add", "-g", "moltbot@\(target)"])
} else {
("npm", ["npm", "install", "-g", "moltbot@\(target)"])
}
statusHandler("Installing moltbot@\(target) via \(label)")
func summarize(_ text: String) -> String? {
let lines = text
.split(whereSeparator: \.isNewline)
.map { $0.trimmingCharacters(in: .whitespacesAndNewlines) }
.filter { !$0.isEmpty }
guard let last = lines.last else { return nil }
let normalized = last.replacingOccurrences(of: "\\s+", with: " ", options: .regularExpression)
return normalized.count > 200 ? String(normalized.prefix(199)) + "" : normalized
}
let response = await ShellExecutor.runDetailed(command: cmd, cwd: nil, env: ["PATH": preferred], timeout: 300)
if response.success {
statusHandler("Installed moltbot@\(target)")
} else {
if response.timedOut {
statusHandler("Install failed: timed out. Check your internet connection and try again.")
return
}
let exit = response.exitCode.map { "exit \($0)" } ?? (response.errorMessage ?? "failed")
let detail = summarize(response.stderr) ?? summarize(response.stdout)
if let detail {
statusHandler("Install failed (\(exit)): \(detail)")
} else {
statusHandler("Install failed (\(exit))")
}
}
}
// MARK: - Internals
private static func readGatewayVersion(binary: String) -> Semver? {
let start = Date()
let process = Process()
process.executableURL = URL(fileURLWithPath: binary)
process.arguments = ["--version"]
process.environment = ["PATH": CommandResolver.preferredPaths().joined(separator: ":")]
let pipe = Pipe()
process.standardOutput = pipe
process.standardError = pipe
do {
let data = try process.runAndReadToEnd(from: pipe)
let elapsedMs = Int(Date().timeIntervalSince(start) * 1000)
if elapsedMs > 500 {
self.logger.warning(
"""
gateway --version slow (\(elapsedMs, privacy: .public)ms) \
bin=\(binary, privacy: .public)
""")
} else {
self.logger.debug(
"""
gateway --version ok (\(elapsedMs, privacy: .public)ms) \
bin=\(binary, privacy: .public)
""")
}
let raw = String(data: data, encoding: .utf8)?
.trimmingCharacters(in: .whitespacesAndNewlines)
return Semver.parse(raw)
} catch {
let elapsedMs = Int(Date().timeIntervalSince(start) * 1000)
self.logger.error(
"""
gateway --version failed (\(elapsedMs, privacy: .public)ms) \
bin=\(binary, privacy: .public) \
err=\(error.localizedDescription, privacy: .public)
""")
return nil
}
}
private static func readLocalGatewayVersion(projectRoot: URL) -> Semver? {
let pkg = projectRoot.appendingPathComponent("package.json")
guard let data = try? Data(contentsOf: pkg) else { return nil }
guard
let json = try? JSONSerialization.jsonObject(with: data) as? [String: Any],
let version = json["version"] as? String
else { return nil }
return Semver.parse(version)
}
}

View File

@@ -0,0 +1,203 @@
import Foundation
enum GatewayLaunchAgentManager {
private static let logger = Logger(subsystem: "bot.molt", category: "gateway.launchd")
private static let disableLaunchAgentMarker = ".clawdbot/disable-launchagent"
private static var disableLaunchAgentMarkerURL: URL {
FileManager().homeDirectoryForCurrentUser
.appendingPathComponent(self.disableLaunchAgentMarker)
}
private static var plistURL: URL {
FileManager().homeDirectoryForCurrentUser
.appendingPathComponent("Library/LaunchAgents/\(gatewayLaunchdLabel).plist")
}
static func isLaunchAgentWriteDisabled() -> Bool {
FileManager().fileExists(atPath: self.disableLaunchAgentMarkerURL.path)
}
static func setLaunchAgentWriteDisabled(_ disabled: Bool) -> String? {
let marker = self.disableLaunchAgentMarkerURL
if disabled {
do {
try FileManager().createDirectory(
at: marker.deletingLastPathComponent(),
withIntermediateDirectories: true)
if !FileManager().fileExists(atPath: marker.path) {
FileManager().createFile(atPath: marker.path, contents: nil)
}
} catch {
return error.localizedDescription
}
return nil
}
if FileManager().fileExists(atPath: marker.path) {
do {
try FileManager().removeItem(at: marker)
} catch {
return error.localizedDescription
}
}
return nil
}
static func isLoaded() async -> Bool {
guard let loaded = await self.readDaemonLoaded() else { return false }
return loaded
}
static func set(enabled: Bool, bundlePath: String, port: Int) async -> String? {
_ = bundlePath
guard !CommandResolver.connectionModeIsRemote() else {
self.logger.info("launchd change skipped (remote mode)")
return nil
}
if enabled, self.isLaunchAgentWriteDisabled() {
self.logger.info("launchd enable skipped (disable marker set)")
return nil
}
if enabled {
self.logger.info("launchd enable requested via CLI port=\(port)")
return await self.runDaemonCommand([
"install",
"--force",
"--port",
"\(port)",
"--runtime",
"node",
])
}
self.logger.info("launchd disable requested via CLI")
return await self.runDaemonCommand(["uninstall"])
}
static func kickstart() async {
_ = await self.runDaemonCommand(["restart"], timeout: 20)
}
static func launchdConfigSnapshot() -> LaunchAgentPlistSnapshot? {
LaunchAgentPlist.snapshot(url: self.plistURL)
}
static func launchdGatewayLogPath() -> String {
let snapshot = self.launchdConfigSnapshot()
if let stdout = snapshot?.stdoutPath?.trimmingCharacters(in: .whitespacesAndNewlines),
!stdout.isEmpty
{
return stdout
}
if let stderr = snapshot?.stderrPath?.trimmingCharacters(in: .whitespacesAndNewlines),
!stderr.isEmpty
{
return stderr
}
return LogLocator.launchdGatewayLogPath
}
}
extension GatewayLaunchAgentManager {
private static func readDaemonLoaded() async -> Bool? {
let result = await self.runDaemonCommandResult(
["status", "--json", "--no-probe"],
timeout: 15,
quiet: true)
guard result.success, let payload = result.payload else { return nil }
guard
let json = try? JSONSerialization.jsonObject(with: payload) as? [String: Any],
let service = json["service"] as? [String: Any],
let loaded = service["loaded"] as? Bool
else {
return nil
}
return loaded
}
private struct CommandResult {
let success: Bool
let payload: Data?
let message: String?
}
private struct ParsedDaemonJson {
let text: String
let object: [String: Any]
}
private static func runDaemonCommand(
_ args: [String],
timeout: Double = 15,
quiet: Bool = false) async -> String?
{
let result = await self.runDaemonCommandResult(args, timeout: timeout, quiet: quiet)
if result.success { return nil }
return result.message ?? "Gateway daemon command failed"
}
private static func runDaemonCommandResult(
_ args: [String],
timeout: Double,
quiet: Bool) async -> CommandResult
{
let command = CommandResolver.clawdbotCommand(
subcommand: "gateway",
extraArgs: self.withJsonFlag(args),
// Launchd management must always run locally, even if remote mode is configured.
configRoot: ["gateway": ["mode": "local"]])
var env = ProcessInfo.processInfo.environment
env["PATH"] = CommandResolver.preferredPaths().joined(separator: ":")
let response = await ShellExecutor.runDetailed(command: command, cwd: nil, env: env, timeout: timeout)
let parsed = self.parseDaemonJson(from: response.stdout) ?? self.parseDaemonJson(from: response.stderr)
let ok = parsed?.object["ok"] as? Bool
let message = (parsed?.object["error"] as? String) ?? (parsed?.object["message"] as? String)
let payload = parsed?.text.data(using: .utf8)
?? (response.stdout.isEmpty ? response.stderr : response.stdout).data(using: .utf8)
let success = ok ?? response.success
if success {
return CommandResult(success: true, payload: payload, message: nil)
}
if quiet {
return CommandResult(success: false, payload: payload, message: message)
}
let detail = message ?? self.summarize(response.stderr) ?? self.summarize(response.stdout)
let exit = response.exitCode.map { "exit \($0)" } ?? (response.errorMessage ?? "failed")
let fullMessage = detail.map { "Gateway daemon command failed (\(exit)): \($0)" }
?? "Gateway daemon command failed (\(exit))"
self.logger.error("\(fullMessage, privacy: .public)")
return CommandResult(success: false, payload: payload, message: detail)
}
private static func withJsonFlag(_ args: [String]) -> [String] {
if args.contains("--json") { return args }
return args + ["--json"]
}
private static func parseDaemonJson(from raw: String) -> ParsedDaemonJson? {
let trimmed = raw.trimmingCharacters(in: .whitespacesAndNewlines)
guard let start = trimmed.firstIndex(of: "{"),
let end = trimmed.lastIndex(of: "}")
else {
return nil
}
let jsonText = String(trimmed[start...end])
guard let data = jsonText.data(using: .utf8) else { return nil }
guard let object = try? JSONSerialization.jsonObject(with: data) as? [String: Any] else { return nil }
return ParsedDaemonJson(text: jsonText, object: object)
}
private static func summarize(_ text: String) -> String? {
let lines = text
.split(whereSeparator: \.isNewline)
.map { $0.trimmingCharacters(in: .whitespacesAndNewlines) }
.filter { !$0.isEmpty }
guard let last = lines.last else { return nil }
let normalized = last.replacingOccurrences(of: "\\s+", with: " ", options: .regularExpression)
return normalized.count > 200 ? String(normalized.prefix(199)) + "" : normalized
}
}

View File

@@ -0,0 +1,432 @@
import Foundation
import Observation
@MainActor
@Observable
final class GatewayProcessManager {
static let shared = GatewayProcessManager()
enum Status: Equatable {
case stopped
case starting
case running(details: String?)
case attachedExisting(details: String?)
case failed(String)
var label: String {
switch self {
case .stopped: return "Stopped"
case .starting: return "Starting…"
case let .running(details):
if let details, !details.isEmpty { return "Running (\(details))" }
return "Running"
case let .attachedExisting(details):
if let details, !details.isEmpty {
return "Using existing gateway (\(details))"
}
return "Using existing gateway"
case let .failed(reason): return "Failed: \(reason)"
}
}
}
private(set) var status: Status = .stopped {
didSet { CanvasManager.shared.refreshDebugStatus() }
}
private(set) var log: String = ""
private(set) var environmentStatus: GatewayEnvironmentStatus = .checking
private(set) var existingGatewayDetails: String?
private(set) var lastFailureReason: String?
private var desiredActive = false
private var environmentRefreshTask: Task<Void, Never>?
private var lastEnvironmentRefresh: Date?
private var logRefreshTask: Task<Void, Never>?
#if DEBUG
private var testingConnection: GatewayConnection?
#endif
private let logger = Logger(subsystem: "bot.molt", category: "gateway.process")
private let logLimit = 20000 // characters to keep in-memory
private let environmentRefreshMinInterval: TimeInterval = 30
private var connection: GatewayConnection {
#if DEBUG
return self.testingConnection ?? .shared
#else
return .shared
#endif
}
func setActive(_ active: Bool) {
// Remote mode should never spawn a local gateway; treat as stopped.
if CommandResolver.connectionModeIsRemote() {
self.desiredActive = false
self.stop()
self.status = .stopped
self.appendLog("[gateway] remote mode active; skipping local gateway\n")
self.logger.info("gateway process skipped: remote mode active")
return
}
self.logger.debug("gateway active requested active=\(active)")
self.desiredActive = active
self.refreshEnvironmentStatus()
if active {
self.startIfNeeded()
} else {
self.stop()
}
}
func ensureLaunchAgentEnabledIfNeeded() async {
guard !CommandResolver.connectionModeIsRemote() else { return }
if GatewayLaunchAgentManager.isLaunchAgentWriteDisabled() {
self.appendLog("[gateway] launchd auto-enable skipped (attach-only)\n")
self.logger.info("gateway launchd auto-enable skipped (disable marker set)")
return
}
let enabled = await GatewayLaunchAgentManager.isLoaded()
guard !enabled else { return }
let bundlePath = Bundle.main.bundleURL.path
let port = GatewayEnvironment.gatewayPort()
self.appendLog("[gateway] auto-enabling launchd job (\(gatewayLaunchdLabel)) on port \(port)\n")
let err = await GatewayLaunchAgentManager.set(enabled: true, bundlePath: bundlePath, port: port)
if let err {
self.appendLog("[gateway] launchd auto-enable failed: \(err)\n")
}
}
func startIfNeeded() {
guard self.desiredActive else { return }
// Do not spawn in remote mode (the gateway should run on the remote host).
guard !CommandResolver.connectionModeIsRemote() else {
self.status = .stopped
return
}
// Many surfaces can call `setActive(true)` in quick succession (startup, Canvas, health checks).
// Avoid spawning multiple concurrent "start" tasks that can thrash launchd and flap the port.
switch self.status {
case .starting, .running, .attachedExisting:
return
case .stopped, .failed:
break
}
self.status = .starting
self.logger.debug("gateway start requested")
// First try to latch onto an already-running gateway to avoid spawning a duplicate.
Task { [weak self] in
guard let self else { return }
if await self.attachExistingGatewayIfAvailable() {
return
}
await self.enableLaunchdGateway()
}
}
func stop() {
self.desiredActive = false
self.existingGatewayDetails = nil
self.lastFailureReason = nil
self.status = .stopped
self.logger.info("gateway stop requested")
if CommandResolver.connectionModeIsRemote() {
return
}
let bundlePath = Bundle.main.bundleURL.path
Task {
_ = await GatewayLaunchAgentManager.set(
enabled: false,
bundlePath: bundlePath,
port: GatewayEnvironment.gatewayPort())
}
}
func clearLastFailure() {
self.lastFailureReason = nil
}
func refreshEnvironmentStatus(force: Bool = false) {
let now = Date()
if !force {
if self.environmentRefreshTask != nil { return }
if let last = self.lastEnvironmentRefresh,
now.timeIntervalSince(last) < self.environmentRefreshMinInterval
{
return
}
}
self.lastEnvironmentRefresh = now
self.environmentRefreshTask = Task { [weak self] in
let status = await Task.detached(priority: .utility) {
GatewayEnvironment.check()
}.value
await MainActor.run {
guard let self else { return }
self.environmentStatus = status
self.environmentRefreshTask = nil
}
}
}
func refreshLog() {
guard self.logRefreshTask == nil else { return }
let path = GatewayLaunchAgentManager.launchdGatewayLogPath()
let limit = self.logLimit
self.logRefreshTask = Task { [weak self] in
let log = await Task.detached(priority: .utility) {
Self.readGatewayLog(path: path, limit: limit)
}.value
await MainActor.run {
guard let self else { return }
if !log.isEmpty {
self.log = log
}
self.logRefreshTask = nil
}
}
}
// MARK: - Internals
/// Attempt to connect to an already-running gateway on the configured port.
/// If successful, mark status as attached and skip spawning a new process.
private func attachExistingGatewayIfAvailable() async -> Bool {
let port = GatewayEnvironment.gatewayPort()
let instance = await PortGuardian.shared.describe(port: port)
let instanceText = instance.map { self.describe(instance: $0) }
let hasListener = instance != nil
let attemptAttach = {
try await self.connection.requestRaw(method: .health, timeoutMs: 2000)
}
for attempt in 0..<(hasListener ? 3 : 1) {
do {
let data = try await attemptAttach()
let snap = decodeHealthSnapshot(from: data)
let details = self.describe(details: instanceText, port: port, snap: snap)
self.existingGatewayDetails = details
self.clearLastFailure()
self.status = .attachedExisting(details: details)
self.appendLog("[gateway] using existing instance: \(details)\n")
self.logger.info("gateway using existing instance details=\(details)")
self.refreshControlChannelIfNeeded(reason: "attach existing")
self.refreshLog()
return true
} catch {
if attempt < 2, hasListener {
try? await Task.sleep(nanoseconds: 250_000_000)
continue
}
if hasListener {
let reason = self.describeAttachFailure(error, port: port, instance: instance)
self.existingGatewayDetails = instanceText
self.status = .failed(reason)
self.lastFailureReason = reason
self.appendLog("[gateway] existing listener on port \(port) but attach failed: \(reason)\n")
self.logger.warning("gateway attach failed reason=\(reason)")
return true
}
// No reachable gateway (and no listener) fall through to spawn.
self.existingGatewayDetails = nil
return false
}
}
self.existingGatewayDetails = nil
return false
}
private func describe(details instance: String?, port: Int, snap: HealthSnapshot?) -> String {
let instanceText = instance ?? "pid unknown"
if let snap {
let order = snap.channelOrder ?? Array(snap.channels.keys)
let linkId = order.first(where: { snap.channels[$0]?.linked == true })
?? order.first(where: { snap.channels[$0]?.linked != nil })
guard let linkId else {
return "port \(port), health probe succeeded, \(instanceText)"
}
let linked = snap.channels[linkId]?.linked ?? false
let authAge = snap.channels[linkId]?.authAgeMs.flatMap(msToAge) ?? "unknown age"
let label =
snap.channelLabels?[linkId] ??
linkId.capitalized
let linkText = linked ? "linked" : "not linked"
return "port \(port), \(label) \(linkText), auth \(authAge), \(instanceText)"
}
return "port \(port), health probe succeeded, \(instanceText)"
}
private func describe(instance: PortGuardian.Descriptor) -> String {
let path = instance.executablePath ?? "path unknown"
return "pid \(instance.pid) \(instance.command) @ \(path)"
}
private func describeAttachFailure(_ error: Error, port: Int, instance: PortGuardian.Descriptor?) -> String {
let ns = error as NSError
let message = ns.localizedDescription.isEmpty ? "unknown error" : ns.localizedDescription
let lower = message.lowercased()
if self.isGatewayAuthFailure(error) {
return """
Gateway on port \(port) rejected auth. Set gateway.auth.token (or CLAWDBOT_GATEWAY_TOKEN) \
to match the running gateway (or clear it on the gateway) and retry.
"""
}
if lower.contains("protocol mismatch") {
return "Gateway on port \(port) is incompatible (protocol mismatch). Update the app/gateway."
}
if lower.contains("unexpected response") || lower.contains("invalid response") {
return "Port \(port) returned non-gateway data; another process is using it."
}
if let instance {
let instanceText = self.describe(instance: instance)
return "Gateway listener found on port \(port) (\(instanceText)) but health check failed: \(message)"
}
return "Gateway listener found on port \(port) but health check failed: \(message)"
}
private func isGatewayAuthFailure(_ error: Error) -> Bool {
if let urlError = error as? URLError, urlError.code == .dataNotAllowed {
return true
}
let ns = error as NSError
if ns.domain == "Gateway", ns.code == 1008 { return true }
let lower = ns.localizedDescription.lowercased()
return lower.contains("unauthorized") || lower.contains("auth")
}
private func enableLaunchdGateway() async {
self.existingGatewayDetails = nil
let resolution = await Task.detached(priority: .utility) {
GatewayEnvironment.resolveGatewayCommand()
}.value
await MainActor.run { self.environmentStatus = resolution.status }
guard resolution.command != nil else {
await MainActor.run {
self.status = .failed(resolution.status.message)
}
self.logger.error("gateway command resolve failed: \(resolution.status.message)")
return
}
if GatewayLaunchAgentManager.isLaunchAgentWriteDisabled() {
let message = "Launchd disabled; start the Gateway manually or disable attach-only."
self.status = .failed(message)
self.lastFailureReason = "launchd disabled"
self.appendLog("[gateway] launchd disabled; skipping auto-start\n")
self.logger.info("gateway launchd enable skipped (disable marker set)")
return
}
let bundlePath = Bundle.main.bundleURL.path
let port = GatewayEnvironment.gatewayPort()
self.appendLog("[gateway] enabling launchd job (\(gatewayLaunchdLabel)) on port \(port)\n")
self.logger.info("gateway enabling launchd port=\(port)")
let err = await GatewayLaunchAgentManager.set(enabled: true, bundlePath: bundlePath, port: port)
if let err {
self.status = .failed(err)
self.lastFailureReason = err
self.logger.error("gateway launchd enable failed: \(err)")
return
}
// Best-effort: wait for the gateway to accept connections.
let deadline = Date().addingTimeInterval(6)
while Date() < deadline {
if !self.desiredActive { return }
do {
_ = try await self.connection.requestRaw(method: .health, timeoutMs: 1500)
let instance = await PortGuardian.shared.describe(port: port)
let details = instance.map { "pid \($0.pid)" }
self.clearLastFailure()
self.status = .running(details: details)
self.logger.info("gateway started details=\(details ?? "ok")")
self.refreshControlChannelIfNeeded(reason: "gateway started")
self.refreshLog()
return
} catch {
try? await Task.sleep(nanoseconds: 400_000_000)
}
}
self.status = .failed("Gateway did not start in time")
self.lastFailureReason = "launchd start timeout"
self.logger.warning("gateway start timed out")
}
private func appendLog(_ chunk: String) {
self.log.append(chunk)
if self.log.count > self.logLimit {
self.log = String(self.log.suffix(self.logLimit))
}
}
private func refreshControlChannelIfNeeded(reason: String) {
switch ControlChannel.shared.state {
case .connected, .connecting:
return
case .disconnected, .degraded:
break
}
self.appendLog("[gateway] refreshing control channel (\(reason))\n")
self.logger.debug("gateway control channel refresh reason=\(reason)")
Task { await ControlChannel.shared.configure() }
}
func waitForGatewayReady(timeout: TimeInterval = 6) async -> Bool {
let deadline = Date().addingTimeInterval(timeout)
while Date() < deadline {
if !self.desiredActive { return false }
do {
_ = try await self.connection.requestRaw(method: .health, timeoutMs: 1500)
self.clearLastFailure()
return true
} catch {
try? await Task.sleep(nanoseconds: 300_000_000)
}
}
self.appendLog("[gateway] readiness wait timed out\n")
self.logger.warning("gateway readiness wait timed out")
return false
}
func clearLog() {
self.log = ""
try? FileManager().removeItem(atPath: GatewayLaunchAgentManager.launchdGatewayLogPath())
self.logger.debug("gateway log cleared")
}
func setProjectRoot(path: String) {
CommandResolver.setProjectRoot(path)
}
func projectRootPath() -> String {
CommandResolver.projectRootPath()
}
private nonisolated static func readGatewayLog(path: String, limit: Int) -> String {
guard FileManager().fileExists(atPath: path) else { return "" }
guard let data = try? Data(contentsOf: URL(fileURLWithPath: path)) else { return "" }
let text = String(data: data, encoding: .utf8) ?? ""
if text.count <= limit { return text }
return String(text.suffix(limit))
}
}
#if DEBUG
extension GatewayProcessManager {
func setTestingConnection(_ connection: GatewayConnection?) {
self.testingConnection = connection
}
func setTestingDesiredActive(_ active: Bool) {
self.desiredActive = active
}
func setTestingLastFailureReason(_ reason: String?) {
self.lastFailureReason = reason
}
}
#endif

View File

@@ -0,0 +1,301 @@
import Foundation
import Network
import Observation
import SwiftUI
struct HealthSnapshot: Codable, Sendable {
struct ChannelSummary: Codable, Sendable {
struct Probe: Codable, Sendable {
struct Bot: Codable, Sendable {
let username: String?
}
struct Webhook: Codable, Sendable {
let url: String?
}
let ok: Bool?
let status: Int?
let error: String?
let elapsedMs: Double?
let bot: Bot?
let webhook: Webhook?
}
let configured: Bool?
let linked: Bool?
let authAgeMs: Double?
let probe: Probe?
let lastProbeAt: Double?
}
struct SessionInfo: Codable, Sendable {
let key: String
let updatedAt: Double?
let age: Double?
}
struct Sessions: Codable, Sendable {
let path: String
let count: Int
let recent: [SessionInfo]
}
let ok: Bool?
let ts: Double
let durationMs: Double
let channels: [String: ChannelSummary]
let channelOrder: [String]?
let channelLabels: [String: String]?
let heartbeatSeconds: Int?
let sessions: Sessions
}
enum HealthState: Equatable {
case unknown
case ok
case linkingNeeded
case degraded(String)
var tint: Color {
switch self {
case .ok: .green
case .linkingNeeded: .red
case .degraded: .orange
case .unknown: .secondary
}
}
}
@MainActor
@Observable
final class HealthStore {
static let shared = HealthStore()
private static let logger = Logger(subsystem: "bot.molt", category: "health")
private(set) var snapshot: HealthSnapshot?
private(set) var lastSuccess: Date?
private(set) var lastError: String?
private(set) var isRefreshing = false
private var loopTask: Task<Void, Never>?
private let refreshInterval: TimeInterval = 60
private init() {
// Avoid background health polling in SwiftUI previews and tests.
if !ProcessInfo.processInfo.isPreview, !ProcessInfo.processInfo.isRunningTests {
self.start()
}
}
// Test-only escape hatch: the HealthStore is a process-wide singleton but
// state derivation is pure from `snapshot` + `lastError`.
func __setSnapshotForTest(_ snapshot: HealthSnapshot?, lastError: String? = nil) {
self.snapshot = snapshot
self.lastError = lastError
}
func start() {
guard self.loopTask == nil else { return }
self.loopTask = Task { [weak self] in
guard let self else { return }
while !Task.isCancelled {
await self.refresh()
try? await Task.sleep(nanoseconds: UInt64(self.refreshInterval * 1_000_000_000))
}
}
}
func stop() {
self.loopTask?.cancel()
self.loopTask = nil
}
func refresh(onDemand: Bool = false) async {
guard !self.isRefreshing else { return }
self.isRefreshing = true
defer { self.isRefreshing = false }
let previousError = self.lastError
do {
let data = try await ControlChannel.shared.health(timeout: 15)
if let decoded = decodeHealthSnapshot(from: data) {
self.snapshot = decoded
self.lastSuccess = Date()
self.lastError = nil
if previousError != nil {
Self.logger.info("health refresh recovered")
}
} else {
self.lastError = "health output not JSON"
if onDemand { self.snapshot = nil }
if previousError != self.lastError {
Self.logger.warning("health refresh failed: output not JSON")
}
}
} catch {
let desc = error.localizedDescription
self.lastError = desc
if onDemand { self.snapshot = nil }
if previousError != desc {
Self.logger.error("health refresh failed \(desc, privacy: .public)")
}
}
}
private static func isChannelHealthy(_ summary: HealthSnapshot.ChannelSummary) -> Bool {
guard summary.configured == true else { return false }
// If probe is missing, treat it as "configured but unknown health" (not a hard fail).
return summary.probe?.ok ?? true
}
private static func describeProbeFailure(_ probe: HealthSnapshot.ChannelSummary.Probe) -> String {
let elapsed = probe.elapsedMs.map { "\(Int($0))ms" }
if let error = probe.error, error.lowercased().contains("timeout") || probe.status == nil {
if let elapsed { return "Health check timed out (\(elapsed))" }
return "Health check timed out"
}
let code = probe.status.map { "status \($0)" } ?? "status unknown"
let reason = probe.error?.isEmpty == false ? probe.error! : "health probe failed"
if let elapsed { return "\(reason) (\(code), \(elapsed))" }
return "\(reason) (\(code))"
}
private func resolveLinkChannel(
_ snap: HealthSnapshot) -> (id: String, summary: HealthSnapshot.ChannelSummary)?
{
let order = snap.channelOrder ?? Array(snap.channels.keys)
for id in order {
if let summary = snap.channels[id], summary.linked == true {
return (id: id, summary: summary)
}
}
for id in order {
if let summary = snap.channels[id], summary.linked != nil {
return (id: id, summary: summary)
}
}
return nil
}
private func resolveFallbackChannel(
_ snap: HealthSnapshot,
excluding id: String?) -> (id: String, summary: HealthSnapshot.ChannelSummary)?
{
let order = snap.channelOrder ?? Array(snap.channels.keys)
for channelId in order {
if channelId == id { continue }
guard let summary = snap.channels[channelId] else { continue }
if Self.isChannelHealthy(summary) {
return (id: channelId, summary: summary)
}
}
return nil
}
var state: HealthState {
if let error = self.lastError, !error.isEmpty {
return .degraded(error)
}
guard let snap = self.snapshot else { return .unknown }
guard let link = self.resolveLinkChannel(snap) else { return .unknown }
if link.summary.linked != true {
// Linking is optional if any other channel is healthy; don't paint the whole app red.
let fallback = self.resolveFallbackChannel(snap, excluding: link.id)
return fallback != nil ? .degraded("Not linked") : .linkingNeeded
}
// A channel can be "linked" but still unhealthy (failed probe / cannot connect).
if let probe = link.summary.probe, probe.ok == false {
return .degraded(Self.describeProbeFailure(probe))
}
return .ok
}
var summaryLine: String {
if self.isRefreshing { return "Health check running…" }
if let error = self.lastError { return "Health check failed: \(error)" }
guard let snap = self.snapshot else { return "Health check pending" }
guard let link = self.resolveLinkChannel(snap) else { return "Health check pending" }
if link.summary.linked != true {
if let fallback = self.resolveFallbackChannel(snap, excluding: link.id) {
let fallbackLabel = snap.channelLabels?[fallback.id] ?? fallback.id.capitalized
let fallbackState = (fallback.summary.probe?.ok ?? true) ? "ok" : "degraded"
return "\(fallbackLabel) \(fallbackState) · Not linked — run moltbot login"
}
return "Not linked — run moltbot login"
}
let auth = link.summary.authAgeMs.map { msToAge($0) } ?? "unknown"
if let probe = link.summary.probe, probe.ok == false {
let status = probe.status.map(String.init) ?? "?"
let suffix = probe.status == nil ? "probe degraded" : "probe degraded · status \(status)"
return "linked · auth \(auth) · \(suffix)"
}
return "linked · auth \(auth)"
}
/// Short, human-friendly detail for the last failure, used in the UI.
var detailLine: String? {
if let error = self.lastError, !error.isEmpty {
let lower = error.lowercased()
if lower.contains("connection refused") {
let port = GatewayEnvironment.gatewayPort()
let host = GatewayConnectivityCoordinator.shared.localEndpointHostLabel ?? "127.0.0.1:\(port)"
return "The gateway control port (\(host)) isnt listening — restart Moltbot to bring it back."
}
if lower.contains("timeout") {
return "Timed out waiting for the control server; the gateway may be crashed or still starting."
}
return error
}
return nil
}
func describeFailure(from snap: HealthSnapshot, fallback: String?) -> String {
if let link = self.resolveLinkChannel(snap), link.summary.linked != true {
return "Not linked — run moltbot login"
}
if let link = self.resolveLinkChannel(snap), let probe = link.summary.probe, probe.ok == false {
return Self.describeProbeFailure(probe)
}
if let fallback, !fallback.isEmpty {
return fallback
}
return "health probe failed"
}
var degradedSummary: String? {
guard case let .degraded(reason) = self.state else { return nil }
if reason == "[object Object]" || reason.trimmingCharacters(in: .whitespacesAndNewlines).isEmpty,
let snap = self.snapshot
{
return self.describeFailure(from: snap, fallback: reason)
}
return reason
}
}
func msToAge(_ ms: Double) -> String {
let minutes = Int(round(ms / 60000))
if minutes < 1 { return "just now" }
if minutes < 60 { return "\(minutes)m" }
let hours = Int(round(Double(minutes) / 60))
if hours < 48 { return "\(hours)h" }
let days = Int(round(Double(hours) / 24))
return "\(days)d"
}
/// Decode a health snapshot, tolerating stray log lines before/after the JSON blob.
func decodeHealthSnapshot(from data: Data) -> HealthSnapshot? {
let decoder = JSONDecoder()
if let snap = try? decoder.decode(HealthSnapshot.self, from: data) {
return snap
}
guard let text = String(data: data, encoding: .utf8) else { return nil }
guard let firstBrace = text.firstIndex(of: "{"), let lastBrace = text.lastIndex(of: "}") else {
return nil
}
let slice = text[firstBrace...lastBrace]
let cleaned = Data(slice.utf8)
return try? decoder.decode(HealthSnapshot.self, from: cleaned)
}

View File

@@ -0,0 +1,394 @@
import MoltbotKit
import MoltbotProtocol
import Cocoa
import Foundation
import Observation
import OSLog
struct InstanceInfo: Identifiable, Codable {
let id: String
let host: String?
let ip: String?
let version: String?
let platform: String?
let deviceFamily: String?
let modelIdentifier: String?
let lastInputSeconds: Int?
let mode: String?
let reason: String?
let text: String
let ts: Double
var ageDescription: String {
let date = Date(timeIntervalSince1970: ts / 1000)
return age(from: date)
}
var lastInputDescription: String {
guard let secs = lastInputSeconds else { return "unknown" }
return "\(secs)s ago"
}
}
@MainActor
@Observable
final class InstancesStore {
static let shared = InstancesStore()
let isPreview: Bool
var instances: [InstanceInfo] = []
var lastError: String?
var statusMessage: String?
var isLoading = false
private let logger = Logger(subsystem: "bot.molt", category: "instances")
private var task: Task<Void, Never>?
private let interval: TimeInterval = 30
private var eventTask: Task<Void, Never>?
private var startCount = 0
private var lastPresenceById: [String: InstanceInfo] = [:]
private var lastLoginNotifiedAtMs: [String: Double] = [:]
private struct PresenceEventPayload: Codable {
let presence: [PresenceEntry]
}
init(isPreview: Bool = false) {
self.isPreview = isPreview
}
func start() {
guard !self.isPreview else { return }
self.startCount += 1
guard self.startCount == 1 else { return }
guard self.task == nil else { return }
self.startGatewaySubscription()
self.task = Task.detached { [weak self] in
guard let self else { return }
await self.refresh()
while !Task.isCancelled {
try? await Task.sleep(nanoseconds: UInt64(self.interval * 1_000_000_000))
await self.refresh()
}
}
}
func stop() {
guard !self.isPreview else { return }
guard self.startCount > 0 else { return }
self.startCount -= 1
guard self.startCount == 0 else { return }
self.task?.cancel()
self.task = nil
self.eventTask?.cancel()
self.eventTask = nil
}
private func startGatewaySubscription() {
self.eventTask?.cancel()
self.eventTask = Task { [weak self] in
guard let self else { return }
let stream = await GatewayConnection.shared.subscribe()
for await push in stream {
if Task.isCancelled { return }
await MainActor.run { [weak self] in
self?.handle(push: push)
}
}
}
}
private func handle(push: GatewayPush) {
switch push {
case let .event(evt) where evt.event == "presence":
if let payload = evt.payload {
self.handlePresenceEventPayload(payload)
}
case .seqGap:
Task { await self.refresh() }
case let .snapshot(hello):
self.applyPresence(hello.snapshot.presence)
default:
break
}
}
func refresh() async {
if self.isLoading { return }
self.statusMessage = nil
self.isLoading = true
defer { self.isLoading = false }
do {
PresenceReporter.shared.sendImmediate(reason: "instances-refresh")
let data = try await ControlChannel.shared.request(method: "system-presence")
self.lastPayload = data
if data.isEmpty {
self.logger.error("instances fetch returned empty payload")
self.instances = [self.localFallbackInstance(reason: "no presence payload")]
self.lastError = nil
self.statusMessage = "No presence payload from gateway; showing local fallback + health probe."
await self.probeHealthIfNeeded(reason: "no payload")
return
}
let decoded = try JSONDecoder().decode([PresenceEntry].self, from: data)
let withIDs = self.normalizePresence(decoded)
if withIDs.isEmpty {
self.instances = [self.localFallbackInstance(reason: "no presence entries")]
self.lastError = nil
self.statusMessage = "Presence list was empty; showing local fallback + health probe."
await self.probeHealthIfNeeded(reason: "empty list")
} else {
self.instances = withIDs
self.lastError = nil
self.statusMessage = nil
}
} catch {
self.logger.error(
"""
instances fetch failed: \(error.localizedDescription, privacy: .public) \
len=\(self.lastPayload?.count ?? 0, privacy: .public) \
utf8=\(self.snippet(self.lastPayload), privacy: .public)
""")
self.instances = [self.localFallbackInstance(reason: "presence decode failed")]
self.lastError = nil
self.statusMessage = "Presence data invalid; showing local fallback + health probe."
await self.probeHealthIfNeeded(reason: "decode failed")
}
}
private func localFallbackInstance(reason: String) -> InstanceInfo {
let host = Host.current().localizedName ?? "this-mac"
let ip = Self.primaryIPv4Address()
let version = Bundle.main.object(forInfoDictionaryKey: "CFBundleShortVersionString") as? String
let osVersion = ProcessInfo.processInfo.operatingSystemVersion
let platform = "macos \(osVersion.majorVersion).\(osVersion.minorVersion).\(osVersion.patchVersion)"
let text = "Local node: \(host)\(ip.map { " (\($0))" } ?? "") · app \(version ?? "dev")"
let ts = Date().timeIntervalSince1970 * 1000
return InstanceInfo(
id: "local-\(host)",
host: host,
ip: ip,
version: version,
platform: platform,
deviceFamily: "Mac",
modelIdentifier: InstanceIdentity.modelIdentifier,
lastInputSeconds: Self.lastInputSeconds(),
mode: "local",
reason: reason,
text: text,
ts: ts)
}
private static func lastInputSeconds() -> Int? {
let anyEvent = CGEventType(rawValue: UInt32.max) ?? .null
let seconds = CGEventSource.secondsSinceLastEventType(.combinedSessionState, eventType: anyEvent)
if seconds.isNaN || seconds.isInfinite || seconds < 0 { return nil }
return Int(seconds.rounded())
}
private static func primaryIPv4Address() -> String? {
var addrList: UnsafeMutablePointer<ifaddrs>?
guard getifaddrs(&addrList) == 0, let first = addrList else { return nil }
defer { freeifaddrs(addrList) }
var fallback: String?
var en0: String?
for ptr in sequence(first: first, next: { $0.pointee.ifa_next }) {
let flags = Int32(ptr.pointee.ifa_flags)
let isUp = (flags & IFF_UP) != 0
let isLoopback = (flags & IFF_LOOPBACK) != 0
let name = String(cString: ptr.pointee.ifa_name)
let family = ptr.pointee.ifa_addr.pointee.sa_family
if !isUp || isLoopback || family != UInt8(AF_INET) { continue }
var addr = ptr.pointee.ifa_addr.pointee
var buffer = [CChar](repeating: 0, count: Int(NI_MAXHOST))
let result = getnameinfo(
&addr,
socklen_t(ptr.pointee.ifa_addr.pointee.sa_len),
&buffer,
socklen_t(buffer.count),
nil,
0,
NI_NUMERICHOST)
guard result == 0 else { continue }
let len = buffer.prefix { $0 != 0 }
let bytes = len.map { UInt8(bitPattern: $0) }
guard let ip = String(bytes: bytes, encoding: .utf8) else { continue }
if name == "en0" { en0 = ip; break }
if fallback == nil { fallback = ip }
}
return en0 ?? fallback
}
// MARK: - Helpers
/// Keep the last raw payload for logging.
private var lastPayload: Data?
private func snippet(_ data: Data?, limit: Int = 256) -> String {
guard let data else { return "<none>" }
if data.isEmpty { return "<empty>" }
let prefix = data.prefix(limit)
if let asString = String(data: prefix, encoding: .utf8) {
return asString.replacingOccurrences(of: "\n", with: " ")
}
return "<\(data.count) bytes non-utf8>"
}
private func probeHealthIfNeeded(reason: String? = nil) async {
do {
let data = try await ControlChannel.shared.health(timeout: 8)
guard let snap = decodeHealthSnapshot(from: data) else { return }
let linkId = snap.channelOrder?.first(where: {
if let summary = snap.channels[$0] { return summary.linked != nil }
return false
}) ?? snap.channels.keys.first(where: {
if let summary = snap.channels[$0] { return summary.linked != nil }
return false
})
let linked = linkId.flatMap { snap.channels[$0]?.linked } ?? false
let linkLabel =
linkId.flatMap { snap.channelLabels?[$0] } ??
linkId?.capitalized ??
"channel"
let entry = InstanceInfo(
id: "health-\(snap.ts)",
host: "gateway (health)",
ip: nil,
version: nil,
platform: nil,
deviceFamily: nil,
modelIdentifier: nil,
lastInputSeconds: nil,
mode: "health",
reason: "health probe",
text: "Health ok · \(linkLabel) linked=\(linked)",
ts: snap.ts)
if !self.instances.contains(where: { $0.id == entry.id }) {
self.instances.insert(entry, at: 0)
}
self.lastError = nil
self.statusMessage =
"Presence unavailable (\(reason ?? "refresh")); showing health probe + local fallback."
} catch {
self.logger.error("instances health probe failed: \(error.localizedDescription, privacy: .public)")
if let reason {
self.statusMessage =
"Presence unavailable (\(reason)), health probe failed: \(error.localizedDescription)"
}
}
}
private func decodeAndApplyPresenceData(_ data: Data) {
do {
let decoded = try JSONDecoder().decode([PresenceEntry].self, from: data)
self.applyPresence(decoded)
} catch {
self.logger.error("presence decode from event failed: \(error.localizedDescription, privacy: .public)")
self.lastError = error.localizedDescription
}
}
func handlePresenceEventPayload(_ payload: MoltbotProtocol.AnyCodable) {
do {
let wrapper = try GatewayPayloadDecoding.decode(payload, as: PresenceEventPayload.self)
self.applyPresence(wrapper.presence)
} catch {
self.logger.error("presence event decode failed: \(error.localizedDescription, privacy: .public)")
self.lastError = error.localizedDescription
}
}
private func normalizePresence(_ entries: [PresenceEntry]) -> [InstanceInfo] {
entries.map { entry -> InstanceInfo in
let key = entry.instanceid ?? entry.host ?? entry.ip ?? entry.text ?? "entry-\(entry.ts)"
return InstanceInfo(
id: key,
host: entry.host,
ip: entry.ip,
version: entry.version,
platform: entry.platform,
deviceFamily: entry.devicefamily,
modelIdentifier: entry.modelidentifier,
lastInputSeconds: entry.lastinputseconds,
mode: entry.mode,
reason: entry.reason,
text: entry.text ?? "Unnamed node",
ts: Double(entry.ts))
}
}
private func applyPresence(_ entries: [PresenceEntry]) {
let withIDs = self.normalizePresence(entries)
self.notifyOnNodeLogin(withIDs)
self.lastPresenceById = Dictionary(uniqueKeysWithValues: withIDs.map { ($0.id, $0) })
self.instances = withIDs
self.statusMessage = nil
self.lastError = nil
}
private func notifyOnNodeLogin(_ instances: [InstanceInfo]) {
for inst in instances {
guard let reason = inst.reason?.trimmingCharacters(in: .whitespacesAndNewlines) else { continue }
guard reason == "node-connected" else { continue }
if let mode = inst.mode?.lowercased(), mode == "local" { continue }
let previous = self.lastPresenceById[inst.id]
if previous?.reason == "node-connected", previous?.ts == inst.ts { continue }
let lastNotified = self.lastLoginNotifiedAtMs[inst.id] ?? 0
if inst.ts <= lastNotified { continue }
self.lastLoginNotifiedAtMs[inst.id] = inst.ts
let name = inst.host?.trimmingCharacters(in: .whitespacesAndNewlines)
let device = name?.isEmpty == false ? name! : inst.id
Task { @MainActor in
_ = await NotificationManager().send(
title: "Node connected",
body: device,
sound: nil,
priority: .active)
}
}
}
}
extension InstancesStore {
static func preview(instances: [InstanceInfo] = [
InstanceInfo(
id: "local",
host: "steipete-mac",
ip: "10.0.0.12",
version: "1.2.3",
platform: "macos 26.2.0",
deviceFamily: "Mac",
modelIdentifier: "Mac16,6",
lastInputSeconds: 12,
mode: "local",
reason: "preview",
text: "Local node: steipete-mac (10.0.0.12) · app 1.2.3",
ts: Date().timeIntervalSince1970 * 1000),
InstanceInfo(
id: "gateway",
host: "gateway",
ip: "100.64.0.2",
version: "1.2.3",
platform: "linux 6.6.0",
deviceFamily: "Linux",
modelIdentifier: "x86_64",
lastInputSeconds: 45,
mode: "remote",
reason: "preview",
text: "Gateway node · tunnel ok",
ts: Date().timeIntervalSince1970 * 1000 - 45000),
]) -> InstancesStore {
let store = InstancesStore(isPreview: true)
store.instances = instances
store.statusMessage = "Preview data"
return store
}
}

View File

@@ -0,0 +1,95 @@
import Foundation
enum LaunchAgentManager {
private static let legacyLaunchdLabels = [
"com.steipete.clawdbot",
"com.clawdbot.mac",
]
private static var plistURL: URL {
FileManager().homeDirectoryForCurrentUser
.appendingPathComponent("Library/LaunchAgents/bot.molt.mac.plist")
}
private static var legacyPlistURLs: [URL] {
self.legacyLaunchdLabels.map { label in
FileManager().homeDirectoryForCurrentUser
.appendingPathComponent("Library/LaunchAgents/\(label).plist")
}
}
static func status() async -> Bool {
guard FileManager().fileExists(atPath: self.plistURL.path) else { return false }
let result = await self.runLaunchctl(["print", "gui/\(getuid())/\(launchdLabel)"])
return result == 0
}
static func set(enabled: Bool, bundlePath: String) async {
if enabled {
for legacyLabel in self.legacyLaunchdLabels {
_ = await self.runLaunchctl(["bootout", "gui/\(getuid())/\(legacyLabel)"])
}
for legacyURL in self.legacyPlistURLs {
try? FileManager().removeItem(at: legacyURL)
}
self.writePlist(bundlePath: bundlePath)
_ = await self.runLaunchctl(["bootout", "gui/\(getuid())/\(launchdLabel)"])
_ = await self.runLaunchctl(["bootstrap", "gui/\(getuid())", self.plistURL.path])
_ = await self.runLaunchctl(["kickstart", "-k", "gui/\(getuid())/\(launchdLabel)"])
} else {
// Disable autostart going forward but leave the current app running.
// bootout would terminate the launchd job immediately (and crash the app if launched via agent).
try? FileManager().removeItem(at: self.plistURL)
}
}
private static func writePlist(bundlePath: String) {
let plist = """
<?xml version="1.0" encoding="UTF-8"?>
<!DOCTYPE plist PUBLIC "-//Apple//DTD PLIST 1.0//EN" "http://www.apple.com/DTDs/PropertyList-1.0.dtd">
<plist version="1.0">
<dict>
<key>Label</key>
<string>bot.molt.mac</string>
<key>ProgramArguments</key>
<array>
<string>\(bundlePath)/Contents/MacOS/Moltbot</string>
</array>
<key>WorkingDirectory</key>
<string>\(FileManager().homeDirectoryForCurrentUser.path)</string>
<key>RunAtLoad</key>
<true/>
<key>KeepAlive</key>
<true/>
<key>EnvironmentVariables</key>
<dict>
<key>PATH</key>
<string>\(CommandResolver.preferredPaths().joined(separator: ":"))</string>
</dict>
<key>StandardOutPath</key>
<string>\(LogLocator.launchdLogPath)</string>
<key>StandardErrorPath</key>
<string>\(LogLocator.launchdLogPath)</string>
</dict>
</plist>
"""
try? plist.write(to: self.plistURL, atomically: true, encoding: .utf8)
}
@discardableResult
private static func runLaunchctl(_ args: [String]) async -> Int32 {
await Task.detached(priority: .utility) { () -> Int32 in
let process = Process()
process.launchPath = "/bin/launchctl"
process.arguments = args
let pipe = Pipe()
process.standardOutput = pipe
process.standardError = pipe
do {
_ = try process.runAndReadToEnd(from: pipe)
return process.terminationStatus
} catch {
return -1
}
}.value
}
}

View File

@@ -0,0 +1,230 @@
import Foundation
@_exported import Logging
import os
import OSLog
typealias Logger = Logging.Logger
enum AppLogSettings {
static let logLevelKey = appLogLevelKey
static func logLevel() -> Logger.Level {
if let raw = UserDefaults.standard.string(forKey: self.logLevelKey),
let level = Logger.Level(rawValue: raw)
{
return level
}
return .info
}
static func setLogLevel(_ level: Logger.Level) {
UserDefaults.standard.set(level.rawValue, forKey: self.logLevelKey)
}
static func fileLoggingEnabled() -> Bool {
UserDefaults.standard.bool(forKey: debugFileLogEnabledKey)
}
}
enum AppLogLevel: String, CaseIterable, Identifiable {
case trace
case debug
case info
case notice
case warning
case error
case critical
static let `default`: AppLogLevel = .info
var id: String { self.rawValue }
var title: String {
switch self {
case .trace: "Trace"
case .debug: "Debug"
case .info: "Info"
case .notice: "Notice"
case .warning: "Warning"
case .error: "Error"
case .critical: "Critical"
}
}
}
enum MoltbotLogging {
private static let labelSeparator = "::"
private static let didBootstrap: Void = {
LoggingSystem.bootstrap { label in
let (subsystem, category) = Self.parseLabel(label)
let osHandler = MoltbotOSLogHandler(subsystem: subsystem, category: category)
let fileHandler = MoltbotFileLogHandler(label: label)
return MultiplexLogHandler([osHandler, fileHandler])
}
}()
static func bootstrapIfNeeded() {
_ = self.didBootstrap
}
static func makeLabel(subsystem: String, category: String) -> String {
"\(subsystem)\(self.labelSeparator)\(category)"
}
static func parseLabel(_ label: String) -> (String, String) {
guard let range = label.range(of: labelSeparator) else {
return ("bot.molt", label)
}
let subsystem = String(label[..<range.lowerBound])
let category = String(label[range.upperBound...])
return (subsystem, category)
}
}
extension Logging.Logger {
init(subsystem: String, category: String) {
MoltbotLogging.bootstrapIfNeeded()
let label = MoltbotLogging.makeLabel(subsystem: subsystem, category: category)
self.init(label: label)
}
}
extension Logger.Message.StringInterpolation {
mutating func appendInterpolation(_ value: some Any, privacy: OSLogPrivacy) {
self.appendInterpolation(String(describing: value))
}
}
struct MoltbotOSLogHandler: LogHandler {
private let osLogger: os.Logger
var metadata: Logger.Metadata = [:]
var logLevel: Logger.Level {
get { AppLogSettings.logLevel() }
set { AppLogSettings.setLogLevel(newValue) }
}
init(subsystem: String, category: String) {
self.osLogger = os.Logger(subsystem: subsystem, category: category)
}
subscript(metadataKey key: String) -> Logger.Metadata.Value? {
get { self.metadata[key] }
set { self.metadata[key] = newValue }
}
func log(
level: Logger.Level,
message: Logger.Message,
metadata: Logger.Metadata?,
source: String,
file: String,
function: String,
line: UInt)
{
let merged = Self.mergeMetadata(self.metadata, metadata)
let rendered = Self.renderMessage(message, metadata: merged)
self.osLogger.log(level: Self.osLogType(for: level), "\(rendered, privacy: .public)")
}
private static func osLogType(for level: Logger.Level) -> OSLogType {
switch level {
case .trace, .debug:
.debug
case .info, .notice:
.info
case .warning:
.default
case .error:
.error
case .critical:
.fault
}
}
private static func mergeMetadata(
_ base: Logger.Metadata,
_ extra: Logger.Metadata?) -> Logger.Metadata
{
guard let extra else { return base }
return base.merging(extra, uniquingKeysWith: { _, new in new })
}
private static func renderMessage(_ message: Logger.Message, metadata: Logger.Metadata) -> String {
guard !metadata.isEmpty else { return message.description }
let meta = metadata
.sorted(by: { $0.key < $1.key })
.map { "\($0.key)=\(self.stringify($0.value))" }
.joined(separator: " ")
return "\(message.description) [\(meta)]"
}
private static func stringify(_ value: Logger.Metadata.Value) -> String {
switch value {
case let .string(text):
text
case let .stringConvertible(value):
String(describing: value)
case let .array(values):
"[" + values.map { self.stringify($0) }.joined(separator: ",") + "]"
case let .dictionary(entries):
"{" + entries.map { "\($0.key)=\(self.stringify($0.value))" }.joined(separator: ",") + "}"
}
}
}
struct MoltbotFileLogHandler: LogHandler {
let label: String
var metadata: Logger.Metadata = [:]
var logLevel: Logger.Level {
get { AppLogSettings.logLevel() }
set { AppLogSettings.setLogLevel(newValue) }
}
subscript(metadataKey key: String) -> Logger.Metadata.Value? {
get { self.metadata[key] }
set { self.metadata[key] = newValue }
}
func log(
level: Logger.Level,
message: Logger.Message,
metadata: Logger.Metadata?,
source: String,
file: String,
function: String,
line: UInt)
{
guard AppLogSettings.fileLoggingEnabled() else { return }
let (subsystem, category) = MoltbotLogging.parseLabel(self.label)
var fields: [String: String] = [
"subsystem": subsystem,
"category": category,
"level": level.rawValue,
"source": source,
"file": file,
"function": function,
"line": "\(line)",
]
let merged = self.metadata.merging(metadata ?? [:], uniquingKeysWith: { _, new in new })
for (key, value) in merged {
fields["meta.\(key)"] = Self.stringify(value)
}
DiagnosticsFileLog.shared.log(category: category, event: message.description, fields: fields)
}
private static func stringify(_ value: Logger.Metadata.Value) -> String {
switch value {
case let .string(text):
text
case let .stringConvertible(value):
String(describing: value)
case let .array(values):
"[" + values.map { self.stringify($0) }.joined(separator: ",") + "]"
case let .dictionary(entries):
"{" + entries.map { "\($0.key)=\(self.stringify($0.value))" }.joined(separator: ",") + "}"
}
}
}

View File

@@ -0,0 +1,471 @@
import AppKit
import Darwin
import Foundation
import MenuBarExtraAccess
import Observation
import OSLog
import Security
import SwiftUI
@main
struct MoltbotApp: App {
@NSApplicationDelegateAdaptor(AppDelegate.self) private var delegate
@State private var state: AppState
private static let logger = Logger(subsystem: "bot.molt", category: "app")
private let gatewayManager = GatewayProcessManager.shared
private let controlChannel = ControlChannel.shared
private let activityStore = WorkActivityStore.shared
private let connectivityCoordinator = GatewayConnectivityCoordinator.shared
@State private var statusItem: NSStatusItem?
@State private var isMenuPresented = false
@State private var isPanelVisible = false
@State private var tailscaleService = TailscaleService.shared
@MainActor
private func updateStatusHighlight() {
self.statusItem?.button?.highlight(self.isPanelVisible)
}
@MainActor
private func updateHoverHUDSuppression() {
HoverHUDController.shared.setSuppressed(self.isMenuPresented || self.isPanelVisible)
}
init() {
MoltbotLogging.bootstrapIfNeeded()
Self.applyAttachOnlyOverrideIfNeeded()
_state = State(initialValue: AppStateStore.shared)
}
var body: some Scene {
MenuBarExtra { MenuContent(state: self.state, updater: self.delegate.updaterController) } label: {
CritterStatusLabel(
isPaused: self.state.isPaused,
isSleeping: self.isGatewaySleeping,
isWorking: self.state.isWorking,
earBoostActive: self.state.earBoostActive,
blinkTick: self.state.blinkTick,
sendCelebrationTick: self.state.sendCelebrationTick,
gatewayStatus: self.gatewayManager.status,
animationsEnabled: self.state.iconAnimationsEnabled && !self.isGatewaySleeping,
iconState: self.effectiveIconState)
}
.menuBarExtraStyle(.menu)
.menuBarExtraAccess(isPresented: self.$isMenuPresented) { item in
self.statusItem = item
MenuSessionsInjector.shared.install(into: item)
self.applyStatusItemAppearance(paused: self.state.isPaused, sleeping: self.isGatewaySleeping)
self.installStatusItemMouseHandler(for: item)
self.updateHoverHUDSuppression()
}
.onChange(of: self.state.isPaused) { _, paused in
self.applyStatusItemAppearance(paused: paused, sleeping: self.isGatewaySleeping)
if self.state.connectionMode == .local {
self.gatewayManager.setActive(!paused)
} else {
self.gatewayManager.stop()
}
}
.onChange(of: self.controlChannel.state) { _, _ in
self.applyStatusItemAppearance(paused: self.state.isPaused, sleeping: self.isGatewaySleeping)
}
.onChange(of: self.gatewayManager.status) { _, _ in
self.applyStatusItemAppearance(paused: self.state.isPaused, sleeping: self.isGatewaySleeping)
}
.onChange(of: self.state.connectionMode) { _, mode in
Task { await ConnectionModeCoordinator.shared.apply(mode: mode, paused: self.state.isPaused) }
CLIInstallPrompter.shared.checkAndPromptIfNeeded(reason: "connection-mode")
}
Settings {
SettingsRootView(state: self.state, updater: self.delegate.updaterController)
.frame(width: SettingsTab.windowWidth, height: SettingsTab.windowHeight, alignment: .topLeading)
.environment(self.tailscaleService)
}
.defaultSize(width: SettingsTab.windowWidth, height: SettingsTab.windowHeight)
.windowResizability(.contentSize)
.onChange(of: self.isMenuPresented) { _, _ in
self.updateStatusHighlight()
self.updateHoverHUDSuppression()
}
}
private func applyStatusItemAppearance(paused: Bool, sleeping: Bool) {
self.statusItem?.button?.appearsDisabled = paused || sleeping
}
private static func applyAttachOnlyOverrideIfNeeded() {
let args = CommandLine.arguments
guard args.contains("--attach-only") || args.contains("--no-launchd") else { return }
if let error = GatewayLaunchAgentManager.setLaunchAgentWriteDisabled(true) {
Self.logger.error("attach-only flag failed: \(error, privacy: .public)")
return
}
Task {
_ = await GatewayLaunchAgentManager.set(
enabled: false,
bundlePath: Bundle.main.bundlePath,
port: GatewayEnvironment.gatewayPort())
}
Self.logger.info("attach-only flag enabled")
}
private var isGatewaySleeping: Bool {
if self.state.isPaused { return false }
switch self.state.connectionMode {
case .unconfigured:
return true
case .remote:
if case .connected = self.controlChannel.state { return false }
return true
case .local:
switch self.gatewayManager.status {
case .running, .starting, .attachedExisting:
if case .connected = self.controlChannel.state { return false }
return true
case .failed, .stopped:
return true
}
}
}
@MainActor
private func installStatusItemMouseHandler(for item: NSStatusItem) {
guard let button = item.button else { return }
if button.subviews.contains(where: { $0 is StatusItemMouseHandlerView }) { return }
WebChatManager.shared.onPanelVisibilityChanged = { [self] visible in
self.isPanelVisible = visible
self.updateStatusHighlight()
self.updateHoverHUDSuppression()
}
CanvasManager.shared.onPanelVisibilityChanged = { [self] visible in
self.state.canvasPanelVisible = visible
}
CanvasManager.shared.defaultAnchorProvider = { [self] in self.statusButtonScreenFrame() }
let handler = StatusItemMouseHandlerView()
handler.translatesAutoresizingMaskIntoConstraints = false
handler.onLeftClick = { [self] in
HoverHUDController.shared.dismiss(reason: "statusItemClick")
self.toggleWebChatPanel()
}
handler.onRightClick = { [self] in
HoverHUDController.shared.dismiss(reason: "statusItemRightClick")
WebChatManager.shared.closePanel()
self.isMenuPresented = true
self.updateStatusHighlight()
}
handler.onHoverChanged = { [self] inside in
HoverHUDController.shared.statusItemHoverChanged(
inside: inside,
anchorProvider: { [self] in self.statusButtonScreenFrame() })
}
button.addSubview(handler)
NSLayoutConstraint.activate([
handler.leadingAnchor.constraint(equalTo: button.leadingAnchor),
handler.trailingAnchor.constraint(equalTo: button.trailingAnchor),
handler.topAnchor.constraint(equalTo: button.topAnchor),
handler.bottomAnchor.constraint(equalTo: button.bottomAnchor),
])
}
@MainActor
private func toggleWebChatPanel() {
HoverHUDController.shared.setSuppressed(true)
self.isMenuPresented = false
Task { @MainActor in
let sessionKey = await WebChatManager.shared.preferredSessionKey()
WebChatManager.shared.togglePanel(
sessionKey: sessionKey,
anchorProvider: { [self] in self.statusButtonScreenFrame() })
}
}
@MainActor
private func statusButtonScreenFrame() -> NSRect? {
guard let button = self.statusItem?.button, let window = button.window else { return nil }
let inWindow = button.convert(button.bounds, to: nil)
return window.convertToScreen(inWindow)
}
private var effectiveIconState: IconState {
let selection = self.state.iconOverride
if selection == .system {
return self.activityStore.iconState
}
let overrideState = selection.toIconState()
switch overrideState {
case let .workingMain(kind): return .overridden(kind)
case let .workingOther(kind): return .overridden(kind)
case .idle: return .idle
case let .overridden(kind): return .overridden(kind)
}
}
}
/// Transparent overlay that intercepts clicks without stealing MenuBarExtra ownership.
private final class StatusItemMouseHandlerView: NSView {
var onLeftClick: (() -> Void)?
var onRightClick: (() -> Void)?
var onHoverChanged: ((Bool) -> Void)?
private var tracking: NSTrackingArea?
override func mouseDown(with event: NSEvent) {
if let onLeftClick {
onLeftClick()
} else {
super.mouseDown(with: event)
}
}
override func rightMouseDown(with event: NSEvent) {
self.onRightClick?()
// Do not call super; menu will be driven by isMenuPresented binding.
}
override func updateTrackingAreas() {
super.updateTrackingAreas()
if let tracking {
self.removeTrackingArea(tracking)
}
let options: NSTrackingArea.Options = [
.mouseEnteredAndExited,
.activeAlways,
.inVisibleRect,
]
let area = NSTrackingArea(rect: self.bounds, options: options, owner: self, userInfo: nil)
self.addTrackingArea(area)
self.tracking = area
}
override func mouseEntered(with event: NSEvent) {
self.onHoverChanged?(true)
}
override func mouseExited(with event: NSEvent) {
self.onHoverChanged?(false)
}
}
@MainActor
final class AppDelegate: NSObject, NSApplicationDelegate {
private var state: AppState?
private let webChatAutoLogger = Logger(subsystem: "bot.molt", category: "Chat")
let updaterController: UpdaterProviding = makeUpdaterController()
func application(_: NSApplication, open urls: [URL]) {
Task { @MainActor in
for url in urls {
await DeepLinkHandler.shared.handle(url: url)
}
}
}
@MainActor
func applicationDidFinishLaunching(_ notification: Notification) {
if self.isDuplicateInstance() {
NSApp.terminate(nil)
return
}
self.state = AppStateStore.shared
AppActivationPolicy.apply(showDockIcon: self.state?.showDockIcon ?? false)
if let state {
Task { await ConnectionModeCoordinator.shared.apply(mode: state.connectionMode, paused: state.isPaused) }
}
TerminationSignalWatcher.shared.start()
NodePairingApprovalPrompter.shared.start()
DevicePairingApprovalPrompter.shared.start()
ExecApprovalsPromptServer.shared.start()
ExecApprovalsGatewayPrompter.shared.start()
MacNodeModeCoordinator.shared.start()
VoiceWakeGlobalSettingsSync.shared.start()
Task { PresenceReporter.shared.start() }
Task { await HealthStore.shared.refresh(onDemand: true) }
Task { await PortGuardian.shared.sweep(mode: AppStateStore.shared.connectionMode) }
Task { await PeekabooBridgeHostCoordinator.shared.setEnabled(AppStateStore.shared.peekabooBridgeEnabled) }
self.scheduleFirstRunOnboardingIfNeeded()
DispatchQueue.main.asyncAfter(deadline: .now() + 1.0) {
CLIInstallPrompter.shared.checkAndPromptIfNeeded(reason: "launch")
}
// Developer/testing helper: auto-open chat when launched with --chat (or legacy --webchat).
if CommandLine.arguments.contains("--chat") || CommandLine.arguments.contains("--webchat") {
self.webChatAutoLogger.debug("Auto-opening chat via CLI flag")
Task { @MainActor in
let sessionKey = await WebChatManager.shared.preferredSessionKey()
WebChatManager.shared.show(sessionKey: sessionKey)
}
}
}
func applicationWillTerminate(_ notification: Notification) {
PresenceReporter.shared.stop()
NodePairingApprovalPrompter.shared.stop()
DevicePairingApprovalPrompter.shared.stop()
ExecApprovalsPromptServer.shared.stop()
ExecApprovalsGatewayPrompter.shared.stop()
MacNodeModeCoordinator.shared.stop()
TerminationSignalWatcher.shared.stop()
VoiceWakeGlobalSettingsSync.shared.stop()
WebChatManager.shared.close()
WebChatManager.shared.resetTunnels()
Task { await RemoteTunnelManager.shared.stopAll() }
Task { await GatewayConnection.shared.shutdown() }
Task { await PeekabooBridgeHostCoordinator.shared.stop() }
}
@MainActor
private func scheduleFirstRunOnboardingIfNeeded() {
let seenVersion = UserDefaults.standard.integer(forKey: onboardingVersionKey)
let shouldShow = seenVersion < currentOnboardingVersion || !AppStateStore.shared.onboardingSeen
guard shouldShow else { return }
DispatchQueue.main.asyncAfter(deadline: .now() + 0.6) {
OnboardingController.shared.show()
}
}
private func isDuplicateInstance() -> Bool {
guard let bundleID = Bundle.main.bundleIdentifier else { return false }
let running = NSWorkspace.shared.runningApplications.filter { $0.bundleIdentifier == bundleID }
return running.count > 1
}
}
// MARK: - Sparkle updater (disabled for unsigned/dev builds)
@MainActor
protocol UpdaterProviding: AnyObject {
var automaticallyChecksForUpdates: Bool { get set }
var automaticallyDownloadsUpdates: Bool { get set }
var isAvailable: Bool { get }
var updateStatus: UpdateStatus { get }
func checkForUpdates(_ sender: Any?)
}
// No-op updater used for debug/dev runs to suppress Sparkle dialogs.
final class DisabledUpdaterController: UpdaterProviding {
var automaticallyChecksForUpdates: Bool = false
var automaticallyDownloadsUpdates: Bool = false
let isAvailable: Bool = false
let updateStatus = UpdateStatus()
func checkForUpdates(_: Any?) {}
}
@MainActor
@Observable
final class UpdateStatus {
static let disabled = UpdateStatus()
var isUpdateReady: Bool
init(isUpdateReady: Bool = false) {
self.isUpdateReady = isUpdateReady
}
}
#if canImport(Sparkle)
import Sparkle
@MainActor
final class SparkleUpdaterController: NSObject, UpdaterProviding {
private lazy var controller = SPUStandardUpdaterController(
startingUpdater: false,
updaterDelegate: self,
userDriverDelegate: nil)
let updateStatus = UpdateStatus()
init(savedAutoUpdate: Bool) {
super.init()
let updater = self.controller.updater
updater.automaticallyChecksForUpdates = savedAutoUpdate
updater.automaticallyDownloadsUpdates = savedAutoUpdate
self.controller.startUpdater()
}
var automaticallyChecksForUpdates: Bool {
get { self.controller.updater.automaticallyChecksForUpdates }
set { self.controller.updater.automaticallyChecksForUpdates = newValue }
}
var automaticallyDownloadsUpdates: Bool {
get { self.controller.updater.automaticallyDownloadsUpdates }
set { self.controller.updater.automaticallyDownloadsUpdates = newValue }
}
var isAvailable: Bool { true }
func checkForUpdates(_ sender: Any?) {
self.controller.checkForUpdates(sender)
}
func updater(_ updater: SPUUpdater, didDownloadUpdate item: SUAppcastItem) {
self.updateStatus.isUpdateReady = true
}
func updater(_ updater: SPUUpdater, failedToDownloadUpdate item: SUAppcastItem, error: Error) {
self.updateStatus.isUpdateReady = false
}
func userDidCancelDownload(_ updater: SPUUpdater) {
self.updateStatus.isUpdateReady = false
}
func updater(
_ updater: SPUUpdater,
userDidMakeChoice choice: SPUUserUpdateChoice,
forUpdate updateItem: SUAppcastItem,
state: SPUUserUpdateState)
{
switch choice {
case .install, .skip:
self.updateStatus.isUpdateReady = false
case .dismiss:
self.updateStatus.isUpdateReady = (state.stage == .downloaded)
@unknown default:
self.updateStatus.isUpdateReady = false
}
}
}
extension SparkleUpdaterController: @preconcurrency SPUUpdaterDelegate {}
private func isDeveloperIDSigned(bundleURL: URL) -> Bool {
var staticCode: SecStaticCode?
guard SecStaticCodeCreateWithPath(bundleURL as CFURL, SecCSFlags(), &staticCode) == errSecSuccess,
let code = staticCode
else { return false }
var infoCF: CFDictionary?
guard SecCodeCopySigningInformation(code, SecCSFlags(rawValue: kSecCSSigningInformation), &infoCF) == errSecSuccess,
let info = infoCF as? [String: Any],
let certs = info[kSecCodeInfoCertificates as String] as? [SecCertificate],
let leaf = certs.first
else {
return false
}
if let summary = SecCertificateCopySubjectSummary(leaf) as String? {
return summary.hasPrefix("Developer ID Application:")
}
return false
}
@MainActor
private func makeUpdaterController() -> UpdaterProviding {
let bundleURL = Bundle.main.bundleURL
let isBundledApp = bundleURL.pathExtension == "app"
guard isBundledApp, isDeveloperIDSigned(bundleURL: bundleURL) else { return DisabledUpdaterController() }
let defaults = UserDefaults.standard
let autoUpdateKey = "autoUpdateEnabled"
// Default to true; honor the user's last choice otherwise.
let savedAutoUpdate = (defaults.object(forKey: autoUpdateKey) as? Bool) ?? true
return SparkleUpdaterController(savedAutoUpdate: savedAutoUpdate)
}
#else
@MainActor
private func makeUpdaterController() -> UpdaterProviding {
DisabledUpdaterController()
}
#endif

View File

@@ -0,0 +1,97 @@
import AVFoundation
import OSLog
import SwiftUI
actor MicLevelMonitor {
private let logger = Logger(subsystem: "bot.molt", category: "voicewake.meter")
private var engine: AVAudioEngine?
private var update: (@Sendable (Double) -> Void)?
private var running = false
private var smoothedLevel: Double = 0
func start(onLevel: @Sendable @escaping (Double) -> Void) async throws {
self.update = onLevel
if self.running { return }
self.logger.info(
"mic level monitor start (\(AudioInputDeviceObserver.defaultInputDeviceSummary(), privacy: .public))")
let engine = AVAudioEngine()
self.engine = engine
let input = engine.inputNode
let format = input.outputFormat(forBus: 0)
guard format.channelCount > 0, format.sampleRate > 0 else {
self.engine = nil
throw NSError(
domain: "MicLevelMonitor",
code: 1,
userInfo: [NSLocalizedDescriptionKey: "No audio input available"])
}
input.removeTap(onBus: 0)
input.installTap(onBus: 0, bufferSize: 512, format: format) { [weak self] buffer, _ in
guard let self else { return }
let level = Self.normalizedLevel(from: buffer)
Task { await self.push(level: level) }
}
engine.prepare()
try engine.start()
self.running = true
}
func stop() {
guard self.running else { return }
if let engine {
engine.inputNode.removeTap(onBus: 0)
engine.stop()
}
self.engine = nil
self.running = false
}
private func push(level: Double) {
self.smoothedLevel = (self.smoothedLevel * 0.45) + (level * 0.55)
guard let update else { return }
let value = self.smoothedLevel
Task { @MainActor in update(value) }
}
private static func normalizedLevel(from buffer: AVAudioPCMBuffer) -> Double {
guard let channel = buffer.floatChannelData?[0] else { return 0 }
let frameCount = Int(buffer.frameLength)
guard frameCount > 0 else { return 0 }
var sum: Float = 0
for i in 0..<frameCount {
let s = channel[i]
sum += s * s
}
let rms = sqrt(sum / Float(frameCount) + 1e-12)
let db = 20 * log10(Double(rms))
let normalized = max(0, min(1, (db + 50) / 50))
return normalized
}
}
struct MicLevelBar: View {
let level: Double
let segments: Int = 12
var body: some View {
HStack(spacing: 3) {
ForEach(0..<self.segments, id: \.self) { idx in
let fill = self.level * Double(self.segments) > Double(idx)
RoundedRectangle(cornerRadius: 2)
.fill(fill ? self.segmentColor(for: idx) : Color.gray.opacity(0.35))
.frame(width: 14, height: 10)
}
}
.padding(4)
.background(
RoundedRectangle(cornerRadius: 6)
.stroke(Color.gray.opacity(0.25), lineWidth: 1))
}
private func segmentColor(for idx: Int) -> Color {
let fraction = Double(idx + 1) / Double(self.segments)
if fraction < 0.65 { return .green }
if fraction < 0.85 { return .yellow }
return .red
}
}

View File

@@ -0,0 +1,156 @@
import Foundation
import JavaScriptCore
enum ModelCatalogLoader {
static var defaultPath: String { self.resolveDefaultPath() }
private static let logger = Logger(subsystem: "bot.molt", category: "models")
private nonisolated static let appSupportDir: URL = {
let base = FileManager().urls(for: .applicationSupportDirectory, in: .userDomainMask).first!
return base.appendingPathComponent("Moltbot", isDirectory: true)
}()
private static var cachePath: URL {
self.appSupportDir.appendingPathComponent("model-catalog/models.generated.js", isDirectory: false)
}
static func load(from path: String) async throws -> [ModelChoice] {
let expanded = (path as NSString).expandingTildeInPath
guard let resolved = self.resolvePath(preferred: expanded) else {
self.logger.error("model catalog load failed: file not found")
throw NSError(
domain: "ModelCatalogLoader",
code: 1,
userInfo: [NSLocalizedDescriptionKey: "Model catalog file not found"])
}
self.logger.debug("model catalog load start file=\(URL(fileURLWithPath: resolved.path).lastPathComponent)")
let source = try String(contentsOfFile: resolved.path, encoding: .utf8)
let sanitized = self.sanitize(source: source)
let ctx = JSContext()
ctx?.exceptionHandler = { _, exception in
if let exception {
self.logger.warning("model catalog JS exception: \(exception)")
}
}
ctx?.evaluateScript(sanitized)
guard let rawModels = ctx?.objectForKeyedSubscript("MODELS")?.toDictionary() as? [String: Any] else {
self.logger.error("model catalog parse failed: MODELS missing")
throw NSError(
domain: "ModelCatalogLoader",
code: 1,
userInfo: [NSLocalizedDescriptionKey: "Failed to parse models.generated.ts"])
}
var choices: [ModelChoice] = []
for (provider, value) in rawModels {
guard let models = value as? [String: Any] else { continue }
for (id, payload) in models {
guard let dict = payload as? [String: Any] else { continue }
let name = dict["name"] as? String ?? id
let ctxWindow = dict["contextWindow"] as? Int
choices.append(ModelChoice(id: id, name: name, provider: provider, contextWindow: ctxWindow))
}
}
let sorted = choices.sorted { lhs, rhs in
if lhs.provider == rhs.provider {
return lhs.name.localizedCaseInsensitiveCompare(rhs.name) == .orderedAscending
}
return lhs.provider.localizedCaseInsensitiveCompare(rhs.provider) == .orderedAscending
}
self.logger.debug("model catalog loaded providers=\(rawModels.count) models=\(sorted.count)")
if resolved.shouldCache {
self.cacheCatalog(sourcePath: resolved.path)
}
return sorted
}
private static func resolveDefaultPath() -> String {
let cache = self.cachePath.path
if FileManager().isReadableFile(atPath: cache) { return cache }
if let bundlePath = self.bundleCatalogPath() { return bundlePath }
if let nodePath = self.nodeModulesCatalogPath() { return nodePath }
return cache
}
private static func resolvePath(preferred: String) -> (path: String, shouldCache: Bool)? {
if FileManager().isReadableFile(atPath: preferred) {
return (preferred, preferred != self.cachePath.path)
}
if let bundlePath = self.bundleCatalogPath(), bundlePath != preferred {
self.logger.warning("model catalog path missing; falling back to bundled catalog")
return (bundlePath, true)
}
let cache = self.cachePath.path
if cache != preferred, FileManager().isReadableFile(atPath: cache) {
self.logger.warning("model catalog path missing; falling back to cached catalog")
return (cache, false)
}
if let nodePath = self.nodeModulesCatalogPath(), nodePath != preferred {
self.logger.warning("model catalog path missing; falling back to node_modules catalog")
return (nodePath, true)
}
return nil
}
private static func bundleCatalogPath() -> String? {
guard let url = Bundle.main.url(forResource: "models.generated", withExtension: "js") else {
return nil
}
return url.path
}
private static func nodeModulesCatalogPath() -> String? {
let roots = [
URL(fileURLWithPath: CommandResolver.projectRootPath()),
URL(fileURLWithPath: FileManager().currentDirectoryPath),
]
for root in roots {
let candidate = root
.appendingPathComponent("node_modules/@mariozechner/pi-ai/dist/models.generated.js")
if FileManager().isReadableFile(atPath: candidate.path) {
return candidate.path
}
}
return nil
}
private static func cacheCatalog(sourcePath: String) {
let destination = self.cachePath
do {
try FileManager().createDirectory(
at: destination.deletingLastPathComponent(),
withIntermediateDirectories: true)
if FileManager().fileExists(atPath: destination.path) {
try FileManager().removeItem(at: destination)
}
try FileManager().copyItem(atPath: sourcePath, toPath: destination.path)
self.logger.debug("model catalog cached file=\(destination.lastPathComponent)")
} catch {
self.logger.warning("model catalog cache failed: \(error.localizedDescription)")
}
}
private static func sanitize(source: String) -> String {
guard let exportRange = source.range(of: "export const MODELS"),
let firstBrace = source[exportRange.upperBound...].firstIndex(of: "{"),
let lastBrace = source.lastIndex(of: "}")
else {
return "var MODELS = {}"
}
var body = String(source[firstBrace...lastBrace])
body = body.replacingOccurrences(
of: #"(?m)\bsatisfies\s+[^,}\n]+"#,
with: "",
options: .regularExpression)
body = body.replacingOccurrences(
of: #"(?m)\bas\s+[^;,\n]+"#,
with: "",
options: .regularExpression)
return "var MODELS = \(body);"
}
}

View File

@@ -0,0 +1,171 @@
import MoltbotKit
import Foundation
import OSLog
@MainActor
final class MacNodeModeCoordinator {
static let shared = MacNodeModeCoordinator()
private let logger = Logger(subsystem: "bot.molt", category: "mac-node")
private var task: Task<Void, Never>?
private let runtime = MacNodeRuntime()
private let session = GatewayNodeSession()
func start() {
guard self.task == nil else { return }
self.task = Task { [weak self] in
await self?.run()
}
}
func stop() {
self.task?.cancel()
self.task = nil
Task { await self.session.disconnect() }
}
func setPreferredGatewayStableID(_ stableID: String?) {
GatewayDiscoveryPreferences.setPreferredStableID(stableID)
Task { await self.session.disconnect() }
}
private func run() async {
var retryDelay: UInt64 = 1_000_000_000
var lastCameraEnabled: Bool?
let defaults = UserDefaults.standard
while !Task.isCancelled {
if await MainActor.run(body: { AppStateStore.shared.isPaused }) {
try? await Task.sleep(nanoseconds: 1_000_000_000)
continue
}
let cameraEnabled = defaults.object(forKey: cameraEnabledKey) as? Bool ?? false
if lastCameraEnabled == nil {
lastCameraEnabled = cameraEnabled
} else if lastCameraEnabled != cameraEnabled {
lastCameraEnabled = cameraEnabled
await self.session.disconnect()
try? await Task.sleep(nanoseconds: 200_000_000)
}
do {
let config = try await GatewayEndpointStore.shared.requireConfig()
let caps = self.currentCaps()
let commands = self.currentCommands(caps: caps)
let permissions = await self.currentPermissions()
let connectOptions = GatewayConnectOptions(
role: "node",
scopes: [],
caps: caps,
commands: commands,
permissions: permissions,
clientId: "moltbot-macos",
clientMode: "node",
clientDisplayName: InstanceIdentity.displayName)
let sessionBox = self.buildSessionBox(url: config.url)
try await self.session.connect(
url: config.url,
token: config.token,
password: config.password,
connectOptions: connectOptions,
sessionBox: sessionBox,
onConnected: { [weak self] in
guard let self else { return }
self.logger.info("mac node connected to gateway")
let mainSessionKey = await GatewayConnection.shared.mainSessionKey()
await self.runtime.updateMainSessionKey(mainSessionKey)
await self.runtime.setEventSender { [weak self] event, payload in
guard let self else { return }
await self.session.sendEvent(event: event, payloadJSON: payload)
}
},
onDisconnected: { [weak self] reason in
guard let self else { return }
await self.runtime.setEventSender(nil)
self.logger.error("mac node disconnected: \(reason, privacy: .public)")
},
onInvoke: { [weak self] req in
guard let self else {
return BridgeInvokeResponse(
id: req.id,
ok: false,
error: MoltbotNodeError(code: .unavailable, message: "UNAVAILABLE: node not ready"))
}
return await self.runtime.handleInvoke(req)
})
retryDelay = 1_000_000_000
try? await Task.sleep(nanoseconds: 1_000_000_000)
} catch {
self.logger.error("mac node gateway connect failed: \(error.localizedDescription, privacy: .public)")
try? await Task.sleep(nanoseconds: min(retryDelay, 10_000_000_000))
retryDelay = min(retryDelay * 2, 10_000_000_000)
}
}
}
private func currentCaps() -> [String] {
var caps: [String] = [MoltbotCapability.canvas.rawValue, MoltbotCapability.screen.rawValue]
if UserDefaults.standard.object(forKey: cameraEnabledKey) as? Bool ?? false {
caps.append(MoltbotCapability.camera.rawValue)
}
let rawLocationMode = UserDefaults.standard.string(forKey: locationModeKey) ?? "off"
if MoltbotLocationMode(rawValue: rawLocationMode) != .off {
caps.append(MoltbotCapability.location.rawValue)
}
return caps
}
private func currentPermissions() async -> [String: Bool] {
let statuses = await PermissionManager.status()
return Dictionary(uniqueKeysWithValues: statuses.map { ($0.key.rawValue, $0.value) })
}
private func currentCommands(caps: [String]) -> [String] {
var commands: [String] = [
MoltbotCanvasCommand.present.rawValue,
MoltbotCanvasCommand.hide.rawValue,
MoltbotCanvasCommand.navigate.rawValue,
MoltbotCanvasCommand.evalJS.rawValue,
MoltbotCanvasCommand.snapshot.rawValue,
MoltbotCanvasA2UICommand.push.rawValue,
MoltbotCanvasA2UICommand.pushJSONL.rawValue,
MoltbotCanvasA2UICommand.reset.rawValue,
MacNodeScreenCommand.record.rawValue,
MoltbotSystemCommand.notify.rawValue,
MoltbotSystemCommand.which.rawValue,
MoltbotSystemCommand.run.rawValue,
MoltbotSystemCommand.execApprovalsGet.rawValue,
MoltbotSystemCommand.execApprovalsSet.rawValue,
]
let capsSet = Set(caps)
if capsSet.contains(MoltbotCapability.camera.rawValue) {
commands.append(MoltbotCameraCommand.list.rawValue)
commands.append(MoltbotCameraCommand.snap.rawValue)
commands.append(MoltbotCameraCommand.clip.rawValue)
}
if capsSet.contains(MoltbotCapability.location.rawValue) {
commands.append(MoltbotLocationCommand.get.rawValue)
}
return commands
}
private func buildSessionBox(url: URL) -> WebSocketSessionBox? {
guard url.scheme?.lowercased() == "wss" else { return nil }
let host = url.host ?? "gateway"
let port = url.port ?? 443
let stableID = "\(host):\(port)"
let stored = GatewayTLSStore.loadFingerprint(stableID: stableID)
let params = GatewayTLSParams(
required: true,
expectedFingerprint: stored,
allowTOFU: stored == nil,
storeKey: stableID)
let session = GatewayTLSPinningSession(params: params)
return WebSocketSessionBox(session: session)
}
}

View File

@@ -0,0 +1,708 @@
import AppKit
import MoltbotDiscovery
import MoltbotIPC
import MoltbotKit
import MoltbotProtocol
import Foundation
import Observation
import OSLog
import UserNotifications
enum NodePairingReconcilePolicy {
static let activeIntervalMs: UInt64 = 15000
static let resyncDelayMs: UInt64 = 250
static func shouldPoll(pendingCount: Int, isPresenting: Bool) -> Bool {
pendingCount > 0 || isPresenting
}
}
@MainActor
@Observable
final class NodePairingApprovalPrompter {
static let shared = NodePairingApprovalPrompter()
private let logger = Logger(subsystem: "bot.molt", category: "node-pairing")
private var task: Task<Void, Never>?
private var reconcileTask: Task<Void, Never>?
private var reconcileOnceTask: Task<Void, Never>?
private var reconcileInFlight = false
private var isStopping = false
private var isPresenting = false
private var queue: [PendingRequest] = []
var pendingCount: Int = 0
var pendingRepairCount: Int = 0
private var activeAlert: NSAlert?
private var activeRequestId: String?
private var alertHostWindow: NSWindow?
private var remoteResolutionsByRequestId: [String: PairingResolution] = [:]
private var autoApproveAttempts: Set<String> = []
private final class AlertHostWindow: NSWindow {
override var canBecomeKey: Bool { true }
override var canBecomeMain: Bool { true }
}
private struct PairingList: Codable {
let pending: [PendingRequest]
let paired: [PairedNode]?
}
private struct PairedNode: Codable, Equatable {
let nodeId: String
let approvedAtMs: Double?
let displayName: String?
let platform: String?
let version: String?
let remoteIp: String?
}
private struct PendingRequest: Codable, Equatable, Identifiable {
let requestId: String
let nodeId: String
let displayName: String?
let platform: String?
let version: String?
let remoteIp: String?
let isRepair: Bool?
let silent: Bool?
let ts: Double
var id: String { self.requestId }
}
private struct PairingResolvedEvent: Codable {
let requestId: String
let nodeId: String
let decision: String
let ts: Double
}
private enum PairingResolution: String {
case approved
case rejected
}
func start() {
guard self.task == nil else { return }
self.isStopping = false
self.reconcileTask?.cancel()
self.reconcileTask = nil
self.task = Task { [weak self] in
guard let self else { return }
_ = try? await GatewayConnection.shared.refresh()
await self.loadPendingRequestsFromGateway()
let stream = await GatewayConnection.shared.subscribe(bufferingNewest: 200)
for await push in stream {
if Task.isCancelled { return }
await MainActor.run { [weak self] in self?.handle(push: push) }
}
}
}
func stop() {
self.isStopping = true
self.endActiveAlert()
self.task?.cancel()
self.task = nil
self.reconcileTask?.cancel()
self.reconcileTask = nil
self.reconcileOnceTask?.cancel()
self.reconcileOnceTask = nil
self.queue.removeAll(keepingCapacity: false)
self.updatePendingCounts()
self.isPresenting = false
self.activeRequestId = nil
self.alertHostWindow?.orderOut(nil)
self.alertHostWindow?.close()
self.alertHostWindow = nil
self.remoteResolutionsByRequestId.removeAll(keepingCapacity: false)
self.autoApproveAttempts.removeAll(keepingCapacity: false)
}
private func loadPendingRequestsFromGateway() async {
// The gateway process may start slightly after the app. Retry a bit so
// pending pairing prompts are still shown on launch.
var delayMs: UInt64 = 200
for attempt in 1...8 {
if Task.isCancelled { return }
do {
let data = try await GatewayConnection.shared.request(
method: "node.pair.list",
params: nil,
timeoutMs: 6000)
guard !data.isEmpty else { return }
let list = try JSONDecoder().decode(PairingList.self, from: data)
let pendingCount = list.pending.count
guard pendingCount > 0 else { return }
self.logger.info(
"loaded \(pendingCount, privacy: .public) pending node pairing request(s) on startup")
await self.apply(list: list)
return
} catch {
if attempt == 8 {
self.logger
.error(
"failed to load pending pairing requests: \(error.localizedDescription, privacy: .public)")
return
}
try? await Task.sleep(nanoseconds: delayMs * 1_000_000)
delayMs = min(delayMs * 2, 2000)
}
}
}
private func reconcileLoop() async {
// Reconcile requests periodically so multiple running apps stay in sync
// (e.g. close dialogs + notify if another machine approves/rejects via app or CLI).
while !Task.isCancelled {
if self.isStopping { break }
if !self.shouldPoll {
self.reconcileTask = nil
return
}
await self.reconcileOnce(timeoutMs: 2500)
try? await Task.sleep(
nanoseconds: NodePairingReconcilePolicy.activeIntervalMs * 1_000_000)
}
self.reconcileTask = nil
}
private func fetchPairingList(timeoutMs: Double) async throws -> PairingList {
let data = try await GatewayConnection.shared.request(
method: "node.pair.list",
params: nil,
timeoutMs: timeoutMs)
return try JSONDecoder().decode(PairingList.self, from: data)
}
private func apply(list: PairingList) async {
if self.isStopping { return }
let pendingById = Dictionary(
uniqueKeysWithValues: list.pending.map { ($0.requestId, $0) })
// Enqueue any missing requests (covers missed pushes while reconnecting).
for req in list.pending.sorted(by: { $0.ts < $1.ts }) {
self.enqueue(req)
}
// Detect resolved requests (approved/rejected elsewhere).
let queued = self.queue
for req in queued {
if pendingById[req.requestId] != nil { continue }
let resolution = self.inferResolution(for: req, list: list)
if self.activeRequestId == req.requestId, self.activeAlert != nil {
self.remoteResolutionsByRequestId[req.requestId] = resolution
self.logger.info(
"""
pairing request resolved elsewhere; closing dialog \
requestId=\(req.requestId, privacy: .public) \
resolution=\(resolution.rawValue, privacy: .public)
""")
self.endActiveAlert()
continue
}
self.logger.info(
"""
pairing request resolved elsewhere requestId=\(req.requestId, privacy: .public) \
resolution=\(resolution.rawValue, privacy: .public)
""")
self.queue.removeAll { $0 == req }
Task { @MainActor in
await self.notify(resolution: resolution, request: req, via: "remote")
}
}
if self.queue.isEmpty {
self.isPresenting = false
}
self.presentNextIfNeeded()
self.updateReconcileLoop()
}
private func inferResolution(for request: PendingRequest, list: PairingList) -> PairingResolution {
let paired = list.paired ?? []
guard let node = paired.first(where: { $0.nodeId == request.nodeId }) else {
return .rejected
}
if request.isRepair == true, let approvedAtMs = node.approvedAtMs {
return approvedAtMs >= request.ts ? .approved : .rejected
}
return .approved
}
private func endActiveAlert() {
guard let alert = self.activeAlert else { return }
if let parent = alert.window.sheetParent {
parent.endSheet(alert.window, returnCode: .abort)
}
self.activeAlert = nil
self.activeRequestId = nil
}
private func requireAlertHostWindow() -> NSWindow {
if let alertHostWindow {
return alertHostWindow
}
let window = AlertHostWindow(
contentRect: NSRect(x: 0, y: 0, width: 520, height: 1),
styleMask: [.borderless],
backing: .buffered,
defer: false)
window.title = ""
window.isReleasedWhenClosed = false
window.level = .floating
window.collectionBehavior = [.canJoinAllSpaces, .fullScreenAuxiliary]
window.isOpaque = false
window.hasShadow = false
window.backgroundColor = .clear
window.ignoresMouseEvents = true
self.alertHostWindow = window
return window
}
private func handle(push: GatewayPush) {
switch push {
case let .event(evt) where evt.event == "node.pair.requested":
guard let payload = evt.payload else { return }
do {
let req = try GatewayPayloadDecoding.decode(payload, as: PendingRequest.self)
self.enqueue(req)
} catch {
self.logger
.error("failed to decode pairing request: \(error.localizedDescription, privacy: .public)")
}
case let .event(evt) where evt.event == "node.pair.resolved":
guard let payload = evt.payload else { return }
do {
let resolved = try GatewayPayloadDecoding.decode(payload, as: PairingResolvedEvent.self)
self.handleResolved(resolved)
} catch {
self.logger
.error(
"failed to decode pairing resolution: \(error.localizedDescription, privacy: .public)")
}
case .snapshot:
self.scheduleReconcileOnce(delayMs: 0)
case .seqGap:
self.scheduleReconcileOnce()
default:
return
}
}
private func enqueue(_ req: PendingRequest) {
if self.queue.contains(req) { return }
self.queue.append(req)
self.updatePendingCounts()
self.presentNextIfNeeded()
self.updateReconcileLoop()
}
private func presentNextIfNeeded() {
guard !self.isStopping else { return }
guard !self.isPresenting else { return }
guard let next = self.queue.first else { return }
self.isPresenting = true
Task { @MainActor [weak self] in
guard let self else { return }
if await self.trySilentApproveIfPossible(next) {
return
}
self.presentAlert(for: next)
}
}
private func presentAlert(for req: PendingRequest) {
self.logger.info("presenting node pairing alert requestId=\(req.requestId, privacy: .public)")
NSApp.activate(ignoringOtherApps: true)
let alert = NSAlert()
alert.alertStyle = .warning
alert.messageText = "Allow node to connect?"
alert.informativeText = Self.describe(req)
// Fail-safe ordering: if the dialog can't be presented, default to "Later".
alert.addButton(withTitle: "Later")
alert.addButton(withTitle: "Approve")
alert.addButton(withTitle: "Reject")
if #available(macOS 11.0, *), alert.buttons.indices.contains(2) {
alert.buttons[2].hasDestructiveAction = true
}
self.activeAlert = alert
self.activeRequestId = req.requestId
let hostWindow = self.requireAlertHostWindow()
// Position the hidden host window so the sheet appears centered on screen.
// (Sheets attach to the top edge of their parent window; if the parent is tiny, it looks "anchored".)
let sheetSize = alert.window.frame.size
if let screen = hostWindow.screen ?? NSScreen.main {
let bounds = screen.visibleFrame
let x = bounds.midX - (sheetSize.width / 2)
let sheetOriginY = bounds.midY - (sheetSize.height / 2)
let hostY = sheetOriginY + sheetSize.height - hostWindow.frame.height
hostWindow.setFrameOrigin(NSPoint(x: x, y: hostY))
} else {
hostWindow.center()
}
hostWindow.makeKeyAndOrderFront(nil)
alert.beginSheetModal(for: hostWindow) { [weak self] response in
Task { @MainActor [weak self] in
guard let self else { return }
self.activeRequestId = nil
self.activeAlert = nil
await self.handleAlertResponse(response, request: req)
hostWindow.orderOut(nil)
}
}
}
private func handleAlertResponse(_ response: NSApplication.ModalResponse, request: PendingRequest) async {
defer {
if self.queue.first == request {
self.queue.removeFirst()
} else {
self.queue.removeAll { $0 == request }
}
self.updatePendingCounts()
self.isPresenting = false
self.presentNextIfNeeded()
self.updateReconcileLoop()
}
// Never approve/reject while shutting down (alerts can get dismissed during app termination).
guard !self.isStopping else { return }
if let resolved = self.remoteResolutionsByRequestId.removeValue(forKey: request.requestId) {
await self.notify(resolution: resolved, request: request, via: "remote")
return
}
switch response {
case .alertFirstButtonReturn:
// Later: leave as pending (CLI can approve/reject). Request will expire on the gateway TTL.
return
case .alertSecondButtonReturn:
_ = await self.approve(requestId: request.requestId)
await self.notify(resolution: .approved, request: request, via: "local")
case .alertThirdButtonReturn:
await self.reject(requestId: request.requestId)
await self.notify(resolution: .rejected, request: request, via: "local")
default:
return
}
}
private func approve(requestId: String) async -> Bool {
do {
try await GatewayConnection.shared.nodePairApprove(requestId: requestId)
self.logger.info("approved node pairing requestId=\(requestId, privacy: .public)")
return true
} catch {
self.logger.error("approve failed requestId=\(requestId, privacy: .public)")
self.logger.error("approve failed: \(error.localizedDescription, privacy: .public)")
return false
}
}
private func reject(requestId: String) async {
do {
try await GatewayConnection.shared.nodePairReject(requestId: requestId)
self.logger.info("rejected node pairing requestId=\(requestId, privacy: .public)")
} catch {
self.logger.error("reject failed requestId=\(requestId, privacy: .public)")
self.logger.error("reject failed: \(error.localizedDescription, privacy: .public)")
}
}
private static func describe(_ req: PendingRequest) -> String {
let name = req.displayName?.trimmingCharacters(in: .whitespacesAndNewlines)
let platform = self.prettyPlatform(req.platform)
let version = req.version?.trimmingCharacters(in: .whitespacesAndNewlines)
let ip = self.prettyIP(req.remoteIp)
var lines: [String] = []
lines.append("Name: \(name?.isEmpty == false ? name! : "Unknown")")
lines.append("Node ID: \(req.nodeId)")
if let platform, !platform.isEmpty { lines.append("Platform: \(platform)") }
if let version, !version.isEmpty { lines.append("App: \(version)") }
if let ip, !ip.isEmpty { lines.append("IP: \(ip)") }
if req.isRepair == true { lines.append("Note: Repair request (token will rotate).") }
return lines.joined(separator: "\n")
}
private static func prettyIP(_ ip: String?) -> String? {
let trimmed = ip?.trimmingCharacters(in: .whitespacesAndNewlines)
guard let trimmed, !trimmed.isEmpty else { return nil }
return trimmed.replacingOccurrences(of: "::ffff:", with: "")
}
private static func prettyPlatform(_ platform: String?) -> String? {
let raw = platform?.trimmingCharacters(in: .whitespacesAndNewlines)
guard let raw, !raw.isEmpty else { return nil }
if raw.lowercased() == "ios" { return "iOS" }
if raw.lowercased() == "macos" { return "macOS" }
return raw
}
private func notify(resolution: PairingResolution, request: PendingRequest, via: String) async {
let center = UNUserNotificationCenter.current()
let settings = await center.notificationSettings()
guard settings.authorizationStatus == .authorized ||
settings.authorizationStatus == .provisional
else {
return
}
let title = resolution == .approved ? "Node pairing approved" : "Node pairing rejected"
let name = request.displayName?.trimmingCharacters(in: .whitespacesAndNewlines)
let device = name?.isEmpty == false ? name! : request.nodeId
let body = "\(device)\n(via \(via))"
_ = await NotificationManager().send(
title: title,
body: body,
sound: nil,
priority: .active)
}
private struct SSHTarget {
let host: String
let port: Int
}
private func trySilentApproveIfPossible(_ req: PendingRequest) async -> Bool {
guard req.silent == true else { return false }
if self.autoApproveAttempts.contains(req.requestId) { return false }
self.autoApproveAttempts.insert(req.requestId)
guard let target = await self.resolveSSHTarget() else {
self.logger.info("silent pairing skipped (no ssh target) requestId=\(req.requestId, privacy: .public)")
return false
}
let user = NSUserName().trimmingCharacters(in: .whitespacesAndNewlines)
guard !user.isEmpty else {
self.logger.info("silent pairing skipped (missing local user) requestId=\(req.requestId, privacy: .public)")
return false
}
let ok = await Self.probeSSH(user: user, host: target.host, port: target.port)
if !ok {
self.logger.info("silent pairing probe failed requestId=\(req.requestId, privacy: .public)")
return false
}
guard await self.approve(requestId: req.requestId) else {
self.logger.info("silent pairing approve failed requestId=\(req.requestId, privacy: .public)")
return false
}
await self.notify(resolution: .approved, request: req, via: "silent-ssh")
if self.queue.first == req {
self.queue.removeFirst()
} else {
self.queue.removeAll { $0 == req }
}
self.updatePendingCounts()
self.isPresenting = false
self.presentNextIfNeeded()
self.updateReconcileLoop()
return true
}
private func resolveSSHTarget() async -> SSHTarget? {
let settings = CommandResolver.connectionSettings()
if !settings.target.isEmpty, let parsed = CommandResolver.parseSSHTarget(settings.target) {
let user = NSUserName().trimmingCharacters(in: .whitespacesAndNewlines)
if let targetUser = parsed.user,
!targetUser.trimmingCharacters(in: .whitespacesAndNewlines).isEmpty,
targetUser != user
{
self.logger.info("silent pairing skipped (ssh user mismatch)")
return nil
}
let host = parsed.host.trimmingCharacters(in: .whitespacesAndNewlines)
guard !host.isEmpty else { return nil }
let port = parsed.port > 0 ? parsed.port : 22
return SSHTarget(host: host, port: port)
}
let model = GatewayDiscoveryModel(localDisplayName: InstanceIdentity.displayName)
model.start()
defer { model.stop() }
let deadline = Date().addingTimeInterval(5.0)
while model.gateways.isEmpty, Date() < deadline {
try? await Task.sleep(nanoseconds: 200_000_000)
}
let preferred = GatewayDiscoveryPreferences.preferredStableID()
let gateway = model.gateways.first { $0.stableID == preferred } ?? model.gateways.first
guard let gateway else { return nil }
let host = (gateway.tailnetDns?.trimmingCharacters(in: .whitespacesAndNewlines).nonEmpty ??
gateway.lanHost?.trimmingCharacters(in: .whitespacesAndNewlines).nonEmpty)
guard let host, !host.isEmpty else { return nil }
let port = gateway.sshPort > 0 ? gateway.sshPort : 22
return SSHTarget(host: host, port: port)
}
private static func probeSSH(user: String, host: String, port: Int) async -> Bool {
await Task.detached(priority: .utility) {
let process = Process()
process.executableURL = URL(fileURLWithPath: "/usr/bin/ssh")
let options = [
"-o", "BatchMode=yes",
"-o", "ConnectTimeout=5",
"-o", "NumberOfPasswordPrompts=0",
"-o", "PreferredAuthentications=publickey",
"-o", "StrictHostKeyChecking=accept-new",
]
guard let target = CommandResolver.makeSSHTarget(user: user, host: host, port: port) else {
return false
}
let args = CommandResolver.sshArguments(
target: target,
identity: "",
options: options,
remoteCommand: ["/usr/bin/true"])
process.arguments = args
let pipe = Pipe()
process.standardOutput = pipe
process.standardError = pipe
do {
_ = try process.runAndReadToEnd(from: pipe)
} catch {
return false
}
return process.terminationStatus == 0
}.value
}
private var shouldPoll: Bool {
NodePairingReconcilePolicy.shouldPoll(
pendingCount: self.queue.count,
isPresenting: self.isPresenting)
}
private func updateReconcileLoop() {
guard !self.isStopping else { return }
if self.shouldPoll {
if self.reconcileTask == nil {
self.reconcileTask = Task { [weak self] in
await self?.reconcileLoop()
}
}
} else {
self.reconcileTask?.cancel()
self.reconcileTask = nil
}
}
private func updatePendingCounts() {
// Keep a cheap observable summary for the menu bar status line.
self.pendingCount = self.queue.count
self.pendingRepairCount = self.queue.count(where: { $0.isRepair == true })
}
private func reconcileOnce(timeoutMs: Double) async {
if self.isStopping { return }
if self.reconcileInFlight { return }
self.reconcileInFlight = true
defer { self.reconcileInFlight = false }
do {
let list = try await self.fetchPairingList(timeoutMs: timeoutMs)
await self.apply(list: list)
} catch {
// best effort: ignore transient connectivity failures
}
}
private func scheduleReconcileOnce(delayMs: UInt64 = NodePairingReconcilePolicy.resyncDelayMs) {
self.reconcileOnceTask?.cancel()
self.reconcileOnceTask = Task { [weak self] in
guard let self else { return }
if delayMs > 0 {
try? await Task.sleep(nanoseconds: delayMs * 1_000_000)
}
await self.reconcileOnce(timeoutMs: 2500)
}
}
private func handleResolved(_ resolved: PairingResolvedEvent) {
let resolution: PairingResolution =
resolved.decision == PairingResolution.approved.rawValue ? .approved : .rejected
if self.activeRequestId == resolved.requestId, self.activeAlert != nil {
self.remoteResolutionsByRequestId[resolved.requestId] = resolution
self.logger.info(
"""
pairing request resolved elsewhere; closing dialog \
requestId=\(resolved.requestId, privacy: .public) \
resolution=\(resolution.rawValue, privacy: .public)
""")
self.endActiveAlert()
return
}
guard let request = self.queue.first(where: { $0.requestId == resolved.requestId }) else {
return
}
self.queue.removeAll { $0.requestId == resolved.requestId }
self.updatePendingCounts()
Task { @MainActor in
await self.notify(resolution: resolution, request: request, via: "remote")
}
if self.queue.isEmpty {
self.isPresenting = false
}
self.presentNextIfNeeded()
self.updateReconcileLoop()
}
}
#if DEBUG
@MainActor
extension NodePairingApprovalPrompter {
static func exerciseForTesting() async {
let prompter = NodePairingApprovalPrompter()
let pending = PendingRequest(
requestId: "req-1",
nodeId: "node-1",
displayName: "Node One",
platform: "macos",
version: "1.0.0",
remoteIp: "127.0.0.1",
isRepair: false,
silent: true,
ts: 1_700_000_000_000)
let paired = PairedNode(
nodeId: "node-1",
approvedAtMs: 1_700_000_000_000,
displayName: "Node One",
platform: "macOS",
version: "1.0.0",
remoteIp: "127.0.0.1")
let list = PairingList(pending: [pending], paired: [paired])
_ = Self.describe(pending)
_ = Self.prettyIP(pending.remoteIp)
_ = Self.prettyPlatform(pending.platform)
_ = prompter.inferResolution(for: pending, list: list)
prompter.queue = [pending]
_ = prompter.shouldPoll
_ = await prompter.trySilentApproveIfPossible(pending)
prompter.queue.removeAll()
}
}
#endif

View File

@@ -0,0 +1,150 @@
import Foundation
import OSLog
enum NodeServiceManager {
private static let logger = Logger(subsystem: "bot.molt", category: "node.service")
static func start() async -> String? {
let result = await self.runServiceCommandResult(
["node", "start"],
timeout: 20,
quiet: false)
if let error = self.errorMessage(from: result, treatNotLoadedAsError: true) {
self.logger.error("node service start failed: \(error, privacy: .public)")
return error
}
return nil
}
static func stop() async -> String? {
let result = await self.runServiceCommandResult(
["node", "stop"],
timeout: 15,
quiet: false)
if let error = self.errorMessage(from: result, treatNotLoadedAsError: false) {
self.logger.error("node service stop failed: \(error, privacy: .public)")
return error
}
return nil
}
}
extension NodeServiceManager {
private struct CommandResult {
let success: Bool
let payload: Data?
let message: String?
let parsed: ParsedServiceJson?
}
private struct ParsedServiceJson {
let text: String
let object: [String: Any]
let ok: Bool?
let result: String?
let message: String?
let error: String?
let hints: [String]
}
private static func runServiceCommandResult(
_ args: [String],
timeout: Double,
quiet: Bool) async -> CommandResult
{
let command = CommandResolver.clawdbotCommand(
subcommand: "service",
extraArgs: self.withJsonFlag(args),
// Service management must always run locally, even if remote mode is configured.
configRoot: ["gateway": ["mode": "local"]])
var env = ProcessInfo.processInfo.environment
env["PATH"] = CommandResolver.preferredPaths().joined(separator: ":")
let response = await ShellExecutor.runDetailed(command: command, cwd: nil, env: env, timeout: timeout)
let parsed = self.parseServiceJson(from: response.stdout) ?? self.parseServiceJson(from: response.stderr)
let ok = parsed?.ok
let message = parsed?.error ?? parsed?.message
let payload = parsed?.text.data(using: .utf8)
?? (response.stdout.isEmpty ? response.stderr : response.stdout).data(using: .utf8)
let success = ok ?? response.success
if success {
return CommandResult(success: true, payload: payload, message: nil, parsed: parsed)
}
if quiet {
return CommandResult(success: false, payload: payload, message: message, parsed: parsed)
}
let detail = message ?? self.summarize(response.stderr) ?? self.summarize(response.stdout)
let exit = response.exitCode.map { "exit \($0)" } ?? (response.errorMessage ?? "failed")
let fullMessage = detail.map { "Node service command failed (\(exit)): \($0)" }
?? "Node service command failed (\(exit))"
self.logger.error("\(fullMessage, privacy: .public)")
return CommandResult(success: false, payload: payload, message: detail, parsed: parsed)
}
private static func errorMessage(from result: CommandResult, treatNotLoadedAsError: Bool) -> String? {
if !result.success {
return result.message ?? "Node service command failed"
}
guard let parsed = result.parsed else { return nil }
if parsed.ok == false {
return self.mergeHints(message: parsed.error ?? parsed.message, hints: parsed.hints)
}
if treatNotLoadedAsError, parsed.result == "not-loaded" {
let base = parsed.message ?? "Node service not loaded."
return self.mergeHints(message: base, hints: parsed.hints)
}
return nil
}
private static func withJsonFlag(_ args: [String]) -> [String] {
if args.contains("--json") { return args }
return args + ["--json"]
}
private static func parseServiceJson(from raw: String) -> ParsedServiceJson? {
let trimmed = raw.trimmingCharacters(in: .whitespacesAndNewlines)
guard let start = trimmed.firstIndex(of: "{"),
let end = trimmed.lastIndex(of: "}")
else {
return nil
}
let jsonText = String(trimmed[start...end])
guard let data = jsonText.data(using: .utf8) else { return nil }
guard let object = try? JSONSerialization.jsonObject(with: data) as? [String: Any] else { return nil }
let ok = object["ok"] as? Bool
let result = object["result"] as? String
let message = object["message"] as? String
let error = object["error"] as? String
let hints = (object["hints"] as? [String]) ?? []
return ParsedServiceJson(
text: jsonText,
object: object,
ok: ok,
result: result,
message: message,
error: error,
hints: hints)
}
private static func mergeHints(message: String?, hints: [String]) -> String? {
let trimmed = message?.trimmingCharacters(in: .whitespacesAndNewlines)
let nonEmpty = trimmed?.isEmpty == false ? trimmed : nil
guard !hints.isEmpty else { return nonEmpty }
let hintText = hints.prefix(2).joined(separator: " · ")
if let nonEmpty {
return "\(nonEmpty) (\(hintText))"
}
return hintText
}
private static func summarize(_ text: String) -> String? {
let lines = text
.split(whereSeparator: \.isNewline)
.map { $0.trimmingCharacters(in: .whitespacesAndNewlines) }
.filter { !$0.isEmpty }
guard let last = lines.last else { return nil }
let normalized = last.replacingOccurrences(of: "\\s+", with: " ", options: .regularExpression)
return normalized.count > 200 ? String(normalized.prefix(199)) + "" : normalized
}
}

View File

@@ -0,0 +1,102 @@
import Foundation
import Observation
import OSLog
struct NodeInfo: Identifiable, Codable {
let nodeId: String
let displayName: String?
let platform: String?
let version: String?
let coreVersion: String?
let uiVersion: String?
let deviceFamily: String?
let modelIdentifier: String?
let remoteIp: String?
let caps: [String]?
let commands: [String]?
let permissions: [String: Bool]?
let paired: Bool?
let connected: Bool?
var id: String { self.nodeId }
var isConnected: Bool { self.connected ?? false }
var isPaired: Bool { self.paired ?? false }
}
private struct NodeListResponse: Codable {
let ts: Double?
let nodes: [NodeInfo]
}
@MainActor
@Observable
final class NodesStore {
static let shared = NodesStore()
var nodes: [NodeInfo] = []
var lastError: String?
var statusMessage: String?
var isLoading = false
private let logger = Logger(subsystem: "bot.molt", category: "nodes")
private var task: Task<Void, Never>?
private let interval: TimeInterval = 30
private var startCount = 0
func start() {
self.startCount += 1
guard self.startCount == 1 else { return }
guard self.task == nil else { return }
self.task = Task.detached { [weak self] in
guard let self else { return }
await self.refresh()
while !Task.isCancelled {
try? await Task.sleep(nanoseconds: UInt64(self.interval * 1_000_000_000))
await self.refresh()
}
}
}
func stop() {
guard self.startCount > 0 else { return }
self.startCount -= 1
guard self.startCount == 0 else { return }
self.task?.cancel()
self.task = nil
}
func refresh() async {
if self.isLoading { return }
self.statusMessage = nil
self.isLoading = true
defer { self.isLoading = false }
do {
let data = try await GatewayConnection.shared.requestRaw(method: "node.list", params: nil, timeoutMs: 8000)
let decoded = try JSONDecoder().decode(NodeListResponse.self, from: data)
self.nodes = decoded.nodes
self.lastError = nil
self.statusMessage = nil
} catch {
if Self.isCancelled(error) {
self.logger.debug("node.list cancelled; keeping last nodes")
if self.nodes.isEmpty {
self.statusMessage = "Refreshing devices…"
}
self.lastError = nil
return
}
self.logger.error("node.list failed \(error.localizedDescription, privacy: .public)")
self.nodes = []
self.lastError = error.localizedDescription
self.statusMessage = nil
}
}
private static func isCancelled(_ error: Error) -> Bool {
if error is CancellationError { return true }
if let urlError = error as? URLError, urlError.code == .cancelled { return true }
let nsError = error as NSError
if nsError.domain == NSURLErrorDomain, nsError.code == NSURLErrorCancelled { return true }
return false
}
}

View File

@@ -0,0 +1,66 @@
import MoltbotIPC
import Foundation
import Security
import UserNotifications
@MainActor
struct NotificationManager {
private let logger = Logger(subsystem: "bot.molt", category: "notifications")
private static let hasTimeSensitiveEntitlement: Bool = {
guard let task = SecTaskCreateFromSelf(nil) else { return false }
let key = "com.apple.developer.usernotifications.time-sensitive" as CFString
guard let val = SecTaskCopyValueForEntitlement(task, key, nil) else { return false }
return (val as? Bool) == true
}()
func send(title: String, body: String, sound: String?, priority: NotificationPriority? = nil) async -> Bool {
let center = UNUserNotificationCenter.current()
let status = await center.notificationSettings()
if status.authorizationStatus == .notDetermined {
let granted = try? await center.requestAuthorization(options: [.alert, .sound, .badge])
if granted != true {
self.logger.warning("notification permission denied (request)")
return false
}
} else if status.authorizationStatus != .authorized {
self.logger.warning("notification permission denied status=\(status.authorizationStatus.rawValue)")
return false
}
let content = UNMutableNotificationContent()
content.title = title
content.body = body
if let soundName = sound, !soundName.isEmpty {
content.sound = UNNotificationSound(named: UNNotificationSoundName(soundName))
}
// Set interruption level based on priority
if let priority {
switch priority {
case .passive:
content.interruptionLevel = .passive
case .active:
content.interruptionLevel = .active
case .timeSensitive:
if Self.hasTimeSensitiveEntitlement {
content.interruptionLevel = .timeSensitive
} else {
self.logger.debug(
"time-sensitive notification requested without entitlement; falling back to active")
content.interruptionLevel = .active
}
}
}
let req = UNNotificationRequest(identifier: UUID().uuidString, content: content, trigger: nil)
do {
try await center.add(req)
self.logger.debug("notification queued")
return true
} catch {
self.logger.error("notification send failed: \(error.localizedDescription)")
return false
}
}
}

View File

@@ -0,0 +1,412 @@
import MoltbotKit
import MoltbotProtocol
import Foundation
import Observation
import OSLog
import SwiftUI
private let onboardingWizardLogger = Logger(subsystem: "bot.molt", category: "onboarding.wizard")
// MARK: - Swift 6 AnyCodable Bridging Helpers
// Bridge between MoltbotProtocol.AnyCodable and the local module to avoid
// Swift 6 strict concurrency type conflicts.
private typealias ProtocolAnyCodable = MoltbotProtocol.AnyCodable
private func bridgeToLocal(_ value: ProtocolAnyCodable) -> AnyCodable {
if let data = try? JSONEncoder().encode(value),
let decoded = try? JSONDecoder().decode(AnyCodable.self, from: data)
{
return decoded
}
return AnyCodable(value.value)
}
private func bridgeToLocal(_ value: ProtocolAnyCodable?) -> AnyCodable? {
value.map(bridgeToLocal)
}
@MainActor
@Observable
final class OnboardingWizardModel {
private(set) var sessionId: String?
private(set) var currentStep: WizardStep?
private(set) var status: String?
private(set) var errorMessage: String?
var isStarting = false
var isSubmitting = false
private var lastStartMode: AppState.ConnectionMode?
private var lastStartWorkspace: String?
private var restartAttempts = 0
private let maxRestartAttempts = 1
var isComplete: Bool { self.status == "done" }
var isRunning: Bool { self.status == "running" }
func reset() {
self.sessionId = nil
self.currentStep = nil
self.status = nil
self.errorMessage = nil
self.isStarting = false
self.isSubmitting = false
self.restartAttempts = 0
self.lastStartMode = nil
self.lastStartWorkspace = nil
}
func startIfNeeded(mode: AppState.ConnectionMode, workspace: String? = nil) async {
guard self.sessionId == nil, !self.isStarting else { return }
guard mode == .local else { return }
if self.shouldSkipWizard() {
self.sessionId = nil
self.currentStep = nil
self.status = "done"
self.errorMessage = nil
return
}
self.isStarting = true
self.errorMessage = nil
self.lastStartMode = mode
self.lastStartWorkspace = workspace
defer { self.isStarting = false }
do {
GatewayProcessManager.shared.setActive(true)
if await GatewayProcessManager.shared.waitForGatewayReady(timeout: 12) == false {
throw NSError(
domain: "Gateway",
code: 1,
userInfo: [NSLocalizedDescriptionKey: "Gateway did not become ready. Check that it is running."])
}
var params: [String: AnyCodable] = ["mode": AnyCodable("local")]
if let workspace, !workspace.isEmpty {
params["workspace"] = AnyCodable(workspace)
}
let res: WizardStartResult = try await GatewayConnection.shared.requestDecoded(
method: .wizardStart,
params: params)
self.applyStartResult(res)
} catch {
self.status = "error"
self.errorMessage = error.localizedDescription
onboardingWizardLogger.error("start failed: \(error.localizedDescription, privacy: .public)")
}
}
func submit(step: WizardStep, value: AnyCodable?) async {
guard let sessionId, !self.isSubmitting else { return }
self.isSubmitting = true
self.errorMessage = nil
defer { self.isSubmitting = false }
do {
var params: [String: AnyCodable] = ["sessionId": AnyCodable(sessionId)]
var answer: [String: AnyCodable] = ["stepId": AnyCodable(step.id)]
if let value {
answer["value"] = value
}
params["answer"] = AnyCodable(answer)
let res: WizardNextResult = try await GatewayConnection.shared.requestDecoded(
method: .wizardNext,
params: params)
self.applyNextResult(res)
} catch {
if self.restartIfSessionLost(error: error) {
return
}
self.status = "error"
self.errorMessage = error.localizedDescription
onboardingWizardLogger.error("submit failed: \(error.localizedDescription, privacy: .public)")
}
}
func cancelIfRunning() async {
guard let sessionId, self.isRunning else { return }
do {
let res: WizardStatusResult = try await GatewayConnection.shared.requestDecoded(
method: .wizardCancel,
params: ["sessionId": AnyCodable(sessionId)])
self.applyStatusResult(res)
} catch {
self.status = "error"
self.errorMessage = error.localizedDescription
onboardingWizardLogger.error("cancel failed: \(error.localizedDescription, privacy: .public)")
}
}
private func applyStartResult(_ res: WizardStartResult) {
self.sessionId = res.sessionid
self.status = wizardStatusString(res.status) ?? (res.done ? "done" : "running")
self.errorMessage = res.error
self.currentStep = decodeWizardStep(res.step)
if self.currentStep == nil, res.step != nil {
onboardingWizardLogger.error("wizard step decode failed")
}
if res.done { self.currentStep = nil }
self.restartAttempts = 0
}
private func applyNextResult(_ res: WizardNextResult) {
let status = wizardStatusString(res.status)
self.status = status ?? self.status
self.errorMessage = res.error
self.currentStep = decodeWizardStep(res.step)
if self.currentStep == nil, res.step != nil {
onboardingWizardLogger.error("wizard step decode failed")
}
if res.done { self.currentStep = nil }
if res.done || status == "done" || status == "cancelled" || status == "error" {
self.sessionId = nil
}
}
private func applyStatusResult(_ res: WizardStatusResult) {
self.status = wizardStatusString(res.status) ?? "unknown"
self.errorMessage = res.error
self.currentStep = nil
self.sessionId = nil
}
private func restartIfSessionLost(error: Error) -> Bool {
guard let gatewayError = error as? GatewayResponseError else { return false }
guard gatewayError.code == ErrorCode.invalidRequest.rawValue else { return false }
let message = gatewayError.message.lowercased()
guard message.contains("wizard not found") || message.contains("wizard not running") else { return false }
guard let mode = self.lastStartMode, self.restartAttempts < self.maxRestartAttempts else {
return false
}
self.restartAttempts += 1
self.sessionId = nil
self.currentStep = nil
self.status = nil
self.errorMessage = "Wizard session lost. Restarting…"
Task { await self.startIfNeeded(mode: mode, workspace: self.lastStartWorkspace) }
return true
}
private func shouldSkipWizard() -> Bool {
let root = MoltbotConfigFile.loadDict()
if let wizard = root["wizard"] as? [String: Any], !wizard.isEmpty {
return true
}
if let gateway = root["gateway"] as? [String: Any],
let auth = gateway["auth"] as? [String: Any]
{
if let mode = auth["mode"] as? String,
!mode.trimmingCharacters(in: .whitespacesAndNewlines).isEmpty
{
return true
}
if let token = auth["token"] as? String,
!token.trimmingCharacters(in: .whitespacesAndNewlines).isEmpty
{
return true
}
if let password = auth["password"] as? String,
!password.trimmingCharacters(in: .whitespacesAndNewlines).isEmpty
{
return true
}
}
return false
}
}
struct OnboardingWizardStepView: View {
let step: WizardStep
let isSubmitting: Bool
let onStepSubmit: (AnyCodable?) -> Void
@State private var textValue: String
@State private var confirmValue: Bool
@State private var selectedIndex: Int
@State private var selectedIndices: Set<Int>
private let optionItems: [WizardOptionItem]
init(step: WizardStep, isSubmitting: Bool, onSubmit: @escaping (AnyCodable?) -> Void) {
self.step = step
self.isSubmitting = isSubmitting
self.onStepSubmit = onSubmit
let options = parseWizardOptions(step.options).enumerated().map { index, option in
WizardOptionItem(index: index, option: option)
}
self.optionItems = options
let initialText = anyCodableString(step.initialvalue)
let initialConfirm = anyCodableBool(step.initialvalue)
let initialIndex = options.firstIndex(where: { anyCodableEqual($0.option.value, step.initialvalue) }) ?? 0
let initialMulti = Set(
options.filter { option in
anyCodableArray(step.initialvalue).contains { anyCodableEqual($0, option.option.value) }
}.map(\.index))
_textValue = State(initialValue: initialText)
_confirmValue = State(initialValue: initialConfirm)
_selectedIndex = State(initialValue: initialIndex)
_selectedIndices = State(initialValue: initialMulti)
}
var body: some View {
VStack(alignment: .leading, spacing: 12) {
if let title = step.title, !title.isEmpty {
Text(title)
.font(.title2.weight(.semibold))
}
if let message = step.message, !message.isEmpty {
Text(message)
.font(.body)
.foregroundStyle(.secondary)
.fixedSize(horizontal: false, vertical: true)
}
switch wizardStepType(self.step) {
case "note":
EmptyView()
case "text":
self.textField
case "confirm":
Toggle("", isOn: self.$confirmValue)
.toggleStyle(.switch)
case "select":
self.selectOptions
case "multiselect":
self.multiselectOptions
case "progress":
ProgressView()
.controlSize(.small)
case "action":
EmptyView()
default:
Text("Unsupported step type")
.foregroundStyle(.secondary)
}
Button(action: self.submit) {
Text(wizardStepType(self.step) == "action" ? "Run" : "Continue")
.frame(minWidth: 120)
}
.buttonStyle(.borderedProminent)
.disabled(self.isSubmitting || self.isBlocked)
}
.frame(maxWidth: .infinity, alignment: .leading)
}
@ViewBuilder
private var textField: some View {
let isSensitive = self.step.sensitive == true
if isSensitive {
SecureField(self.step.placeholder ?? "", text: self.$textValue)
.textFieldStyle(.roundedBorder)
.frame(maxWidth: 360)
} else {
TextField(self.step.placeholder ?? "", text: self.$textValue)
.textFieldStyle(.roundedBorder)
.frame(maxWidth: 360)
}
}
private var selectOptions: some View {
VStack(alignment: .leading, spacing: 8) {
ForEach(self.optionItems, id: \.index) { item in
self.selectOptionRow(item)
}
}
}
private var multiselectOptions: some View {
VStack(alignment: .leading, spacing: 8) {
ForEach(self.optionItems, id: \.index) { item in
self.multiselectOptionRow(item)
}
}
}
private func selectOptionRow(_ item: WizardOptionItem) -> some View {
Button {
self.selectedIndex = item.index
} label: {
HStack(alignment: .top, spacing: 8) {
Image(systemName: self.selectedIndex == item.index ? "largecircle.fill.circle" : "circle")
.foregroundStyle(Color.accentColor)
VStack(alignment: .leading, spacing: 2) {
Text(item.option.label)
.foregroundStyle(.primary)
if let hint = item.option.hint, !hint.isEmpty {
Text(hint)
.font(.caption)
.foregroundStyle(.secondary)
}
}
}
}
.buttonStyle(.plain)
}
private func multiselectOptionRow(_ item: WizardOptionItem) -> some View {
Toggle(isOn: self.bindingForOption(item)) {
VStack(alignment: .leading, spacing: 2) {
Text(item.option.label)
if let hint = item.option.hint, !hint.isEmpty {
Text(hint)
.font(.caption)
.foregroundStyle(.secondary)
}
}
}
}
private func bindingForOption(_ item: WizardOptionItem) -> Binding<Bool> {
Binding(get: {
self.selectedIndices.contains(item.index)
}, set: { newValue in
if newValue {
self.selectedIndices.insert(item.index)
} else {
self.selectedIndices.remove(item.index)
}
})
}
private var isBlocked: Bool {
let type = wizardStepType(step)
if type == "select" { return self.optionItems.isEmpty }
if type == "multiselect" { return self.optionItems.isEmpty }
return false
}
private func submit() {
switch wizardStepType(self.step) {
case "note", "progress":
self.onStepSubmit(nil)
case "text":
self.onStepSubmit(AnyCodable(self.textValue))
case "confirm":
self.onStepSubmit(AnyCodable(self.confirmValue))
case "select":
guard self.optionItems.indices.contains(self.selectedIndex) else {
self.onStepSubmit(nil)
return
}
let option = self.optionItems[self.selectedIndex].option
self.onStepSubmit(bridgeToLocal(option.value) ?? AnyCodable(option.label))
case "multiselect":
let values = self.optionItems
.filter { self.selectedIndices.contains($0.index) }
.map { bridgeToLocal($0.option.value) ?? AnyCodable($0.option.label) }
self.onStepSubmit(AnyCodable(values))
case "action":
self.onStepSubmit(AnyCodable(true))
default:
self.onStepSubmit(nil)
}
}
}
private struct WizardOptionItem: Identifiable {
let index: Int
let option: WizardOption
var id: Int { self.index }
}

View File

@@ -0,0 +1,130 @@
import Foundation
import os
import PeekabooAutomationKit
import PeekabooBridge
import PeekabooFoundation
import Security
@MainActor
final class PeekabooBridgeHostCoordinator {
static let shared = PeekabooBridgeHostCoordinator()
private let logger = Logger(subsystem: "bot.molt", category: "PeekabooBridge")
private var host: PeekabooBridgeHost?
private var services: MoltbotPeekabooBridgeServices?
func setEnabled(_ enabled: Bool) async {
if enabled {
await self.startIfNeeded()
} else {
await self.stop()
}
}
func stop() async {
guard let host else { return }
await host.stop()
self.host = nil
self.services = nil
self.logger.info("PeekabooBridge host stopped")
}
private func startIfNeeded() async {
guard self.host == nil else { return }
var allowlistedTeamIDs: Set<String> = ["Y5PE65HELJ"]
if let teamID = Self.currentTeamID() {
allowlistedTeamIDs.insert(teamID)
}
let allowlistedBundles: Set<String> = []
let services = MoltbotPeekabooBridgeServices()
let server = PeekabooBridgeServer(
services: services,
hostKind: .gui,
allowlistedTeams: allowlistedTeamIDs,
allowlistedBundles: allowlistedBundles)
let host = PeekabooBridgeHost(
socketPath: PeekabooBridgeConstants.clawdbotSocketPath,
server: server,
allowedTeamIDs: allowlistedTeamIDs,
requestTimeoutSec: 10)
self.services = services
self.host = host
await host.start()
self.logger
.info("PeekabooBridge host started at \(PeekabooBridgeConstants.clawdbotSocketPath, privacy: .public)")
}
private static func currentTeamID() -> String? {
var code: SecCode?
guard SecCodeCopySelf(SecCSFlags(), &code) == errSecSuccess,
let code
else {
return nil
}
var staticCode: SecStaticCode?
guard SecCodeCopyStaticCode(code, SecCSFlags(), &staticCode) == errSecSuccess,
let staticCode
else {
return nil
}
var infoCF: CFDictionary?
guard SecCodeCopySigningInformation(
staticCode,
SecCSFlags(rawValue: kSecCSSigningInformation),
&infoCF) == errSecSuccess,
let info = infoCF as? [String: Any]
else {
return nil
}
return info[kSecCodeInfoTeamIdentifier as String] as? String
}
}
@MainActor
private final class MoltbotPeekabooBridgeServices: PeekabooBridgeServiceProviding {
let permissions: PermissionsService
let screenCapture: any ScreenCaptureServiceProtocol
let automation: any UIAutomationServiceProtocol
let windows: any WindowManagementServiceProtocol
let applications: any ApplicationServiceProtocol
let menu: any MenuServiceProtocol
let dock: any DockServiceProtocol
let dialogs: any DialogServiceProtocol
let snapshots: any SnapshotManagerProtocol
init() {
let logging = LoggingService(subsystem: "bot.molt.peekaboo")
let feedbackClient: any AutomationFeedbackClient = NoopAutomationFeedbackClient()
let snapshots = InMemorySnapshotManager(options: .init(
snapshotValidityWindow: 600,
maxSnapshots: 50,
deleteArtifactsOnCleanup: false))
let applications = ApplicationService(feedbackClient: feedbackClient)
let screenCapture = ScreenCaptureService(loggingService: logging)
self.permissions = PermissionsService()
self.snapshots = snapshots
self.applications = applications
self.screenCapture = screenCapture
self.automation = UIAutomationService(
snapshotManager: snapshots,
loggingService: logging,
searchPolicy: .balanced,
feedbackClient: feedbackClient)
self.windows = WindowManagementService(applicationService: applications, feedbackClient: feedbackClient)
self.menu = MenuService(applicationService: applications, feedbackClient: feedbackClient)
self.dock = DockService(feedbackClient: feedbackClient)
self.dialogs = DialogService(feedbackClient: feedbackClient)
}
}

View File

@@ -0,0 +1,506 @@
import AppKit
import ApplicationServices
import AVFoundation
import MoltbotIPC
import CoreGraphics
import CoreLocation
import Foundation
import Observation
import Speech
import UserNotifications
enum PermissionManager {
static func isLocationAuthorized(status: CLAuthorizationStatus, requireAlways: Bool) -> Bool {
if requireAlways { return status == .authorizedAlways }
switch status {
case .authorizedAlways, .authorizedWhenInUse:
return true
case .authorized: // deprecated, but still shows up on some macOS versions
return true
default:
return false
}
}
static func ensure(_ caps: [Capability], interactive: Bool) async -> [Capability: Bool] {
var results: [Capability: Bool] = [:]
for cap in caps {
results[cap] = await self.ensureCapability(cap, interactive: interactive)
}
return results
}
private static func ensureCapability(_ cap: Capability, interactive: Bool) async -> Bool {
switch cap {
case .notifications:
await self.ensureNotifications(interactive: interactive)
case .appleScript:
await self.ensureAppleScript(interactive: interactive)
case .accessibility:
await self.ensureAccessibility(interactive: interactive)
case .screenRecording:
await self.ensureScreenRecording(interactive: interactive)
case .microphone:
await self.ensureMicrophone(interactive: interactive)
case .speechRecognition:
await self.ensureSpeechRecognition(interactive: interactive)
case .camera:
await self.ensureCamera(interactive: interactive)
case .location:
await self.ensureLocation(interactive: interactive)
}
}
private static func ensureNotifications(interactive: Bool) async -> Bool {
let center = UNUserNotificationCenter.current()
let settings = await center.notificationSettings()
switch settings.authorizationStatus {
case .authorized, .provisional, .ephemeral:
return true
case .notDetermined:
guard interactive else { return false }
let granted = await (try? center.requestAuthorization(options: [.alert, .sound, .badge])) ?? false
let updated = await center.notificationSettings()
return granted &&
(updated.authorizationStatus == .authorized || updated.authorizationStatus == .provisional)
case .denied:
if interactive {
NotificationPermissionHelper.openSettings()
}
return false
@unknown default:
return false
}
}
private static func ensureAppleScript(interactive: Bool) async -> Bool {
let granted = await MainActor.run { AppleScriptPermission.isAuthorized() }
if interactive, !granted {
await AppleScriptPermission.requestAuthorization()
}
return await MainActor.run { AppleScriptPermission.isAuthorized() }
}
private static func ensureAccessibility(interactive: Bool) async -> Bool {
let trusted = await MainActor.run { AXIsProcessTrusted() }
if interactive, !trusted {
await MainActor.run {
let opts: NSDictionary = ["AXTrustedCheckOptionPrompt": true]
_ = AXIsProcessTrustedWithOptions(opts)
}
}
return await MainActor.run { AXIsProcessTrusted() }
}
private static func ensureScreenRecording(interactive: Bool) async -> Bool {
let granted = ScreenRecordingProbe.isAuthorized()
if interactive, !granted {
await ScreenRecordingProbe.requestAuthorization()
}
return ScreenRecordingProbe.isAuthorized()
}
private static func ensureMicrophone(interactive: Bool) async -> Bool {
let status = AVCaptureDevice.authorizationStatus(for: .audio)
switch status {
case .authorized:
return true
case .notDetermined:
guard interactive else { return false }
return await AVCaptureDevice.requestAccess(for: .audio)
case .denied, .restricted:
if interactive {
MicrophonePermissionHelper.openSettings()
}
return false
@unknown default:
return false
}
}
private static func ensureSpeechRecognition(interactive: Bool) async -> Bool {
let status = SFSpeechRecognizer.authorizationStatus()
if status == .notDetermined, interactive {
await withUnsafeContinuation { (cont: UnsafeContinuation<Void, Never>) in
SFSpeechRecognizer.requestAuthorization { _ in
DispatchQueue.main.async { cont.resume() }
}
}
}
return SFSpeechRecognizer.authorizationStatus() == .authorized
}
private static func ensureCamera(interactive: Bool) async -> Bool {
let status = AVCaptureDevice.authorizationStatus(for: .video)
switch status {
case .authorized:
return true
case .notDetermined:
guard interactive else { return false }
return await AVCaptureDevice.requestAccess(for: .video)
case .denied, .restricted:
if interactive {
CameraPermissionHelper.openSettings()
}
return false
@unknown default:
return false
}
}
private static func ensureLocation(interactive: Bool) async -> Bool {
guard CLLocationManager.locationServicesEnabled() else {
if interactive {
await MainActor.run { LocationPermissionHelper.openSettings() }
}
return false
}
let status = CLLocationManager().authorizationStatus
switch status {
case .authorizedAlways, .authorizedWhenInUse, .authorized:
return true
case .notDetermined:
guard interactive else { return false }
let updated = await LocationPermissionRequester.shared.request(always: false)
return self.isLocationAuthorized(status: updated, requireAlways: false)
case .denied, .restricted:
if interactive {
await MainActor.run { LocationPermissionHelper.openSettings() }
}
return false
@unknown default:
return false
}
}
static func voiceWakePermissionsGranted() -> Bool {
let mic = AVCaptureDevice.authorizationStatus(for: .audio) == .authorized
let speech = SFSpeechRecognizer.authorizationStatus() == .authorized
return mic && speech
}
static func ensureVoiceWakePermissions(interactive: Bool) async -> Bool {
let results = await self.ensure([.microphone, .speechRecognition], interactive: interactive)
return results[.microphone] == true && results[.speechRecognition] == true
}
static func status(_ caps: [Capability] = Capability.allCases) async -> [Capability: Bool] {
var results: [Capability: Bool] = [:]
for cap in caps {
switch cap {
case .notifications:
let center = UNUserNotificationCenter.current()
let settings = await center.notificationSettings()
results[cap] = settings.authorizationStatus == .authorized
|| settings.authorizationStatus == .provisional
case .appleScript:
results[cap] = await MainActor.run { AppleScriptPermission.isAuthorized() }
case .accessibility:
results[cap] = await MainActor.run { AXIsProcessTrusted() }
case .screenRecording:
if #available(macOS 10.15, *) {
results[cap] = CGPreflightScreenCaptureAccess()
} else {
results[cap] = true
}
case .microphone:
results[cap] = AVCaptureDevice.authorizationStatus(for: .audio) == .authorized
case .speechRecognition:
results[cap] = SFSpeechRecognizer.authorizationStatus() == .authorized
case .camera:
results[cap] = AVCaptureDevice.authorizationStatus(for: .video) == .authorized
case .location:
let status = CLLocationManager().authorizationStatus
results[cap] = CLLocationManager.locationServicesEnabled()
&& self.isLocationAuthorized(status: status, requireAlways: false)
}
}
return results
}
}
enum NotificationPermissionHelper {
static func openSettings() {
let candidates = [
"x-apple.systempreferences:com.apple.Notifications-Settings.extension",
"x-apple.systempreferences:com.apple.preference.notifications",
]
for candidate in candidates {
if let url = URL(string: candidate), NSWorkspace.shared.open(url) {
return
}
}
}
}
enum MicrophonePermissionHelper {
static func openSettings() {
let candidates = [
"x-apple.systempreferences:com.apple.preference.security?Privacy_Microphone",
"x-apple.systempreferences:com.apple.preference.security",
]
for candidate in candidates {
if let url = URL(string: candidate), NSWorkspace.shared.open(url) {
return
}
}
}
}
enum CameraPermissionHelper {
static func openSettings() {
let candidates = [
"x-apple.systempreferences:com.apple.preference.security?Privacy_Camera",
"x-apple.systempreferences:com.apple.preference.security",
]
for candidate in candidates {
if let url = URL(string: candidate), NSWorkspace.shared.open(url) {
return
}
}
}
}
enum LocationPermissionHelper {
static func openSettings() {
let candidates = [
"x-apple.systempreferences:com.apple.preference.security?Privacy_LocationServices",
"x-apple.systempreferences:com.apple.preference.security",
]
for candidate in candidates {
if let url = URL(string: candidate), NSWorkspace.shared.open(url) {
return
}
}
}
}
@MainActor
final class LocationPermissionRequester: NSObject, CLLocationManagerDelegate {
static let shared = LocationPermissionRequester()
private let manager = CLLocationManager()
private var continuation: CheckedContinuation<CLAuthorizationStatus, Never>?
private var timeoutTask: Task<Void, Never>?
override init() {
super.init()
self.manager.delegate = self
}
func request(always: Bool) async -> CLAuthorizationStatus {
let current = self.manager.authorizationStatus
if PermissionManager.isLocationAuthorized(status: current, requireAlways: always) {
return current
}
return await withCheckedContinuation { cont in
self.continuation = cont
self.timeoutTask?.cancel()
self.timeoutTask = Task { [weak self] in
try? await Task.sleep(nanoseconds: 3_000_000_000)
await MainActor.run { [weak self] in
guard let self else { return }
guard self.continuation != nil else { return }
LocationPermissionHelper.openSettings()
self.finish(status: self.manager.authorizationStatus)
}
}
if always {
self.manager.requestAlwaysAuthorization()
} else {
self.manager.requestWhenInUseAuthorization()
}
// On macOS, requesting an actual fix makes the prompt more reliable.
self.manager.requestLocation()
}
}
private func finish(status: CLAuthorizationStatus) {
self.timeoutTask?.cancel()
self.timeoutTask = nil
guard let cont = self.continuation else { return }
self.continuation = nil
cont.resume(returning: status)
}
// nonisolated for Swift 6 strict concurrency compatibility
nonisolated func locationManagerDidChangeAuthorization(_ manager: CLLocationManager) {
let status = manager.authorizationStatus
Task { @MainActor in
self.finish(status: status)
}
}
// Legacy callback (still used on some macOS versions / configurations).
nonisolated func locationManager(
_ manager: CLLocationManager,
didChangeAuthorization status: CLAuthorizationStatus)
{
Task { @MainActor in
self.finish(status: status)
}
}
nonisolated func locationManager(_ manager: CLLocationManager, didFailWithError error: Error) {
let status = manager.authorizationStatus
Task { @MainActor in
if status == .denied || status == .restricted {
LocationPermissionHelper.openSettings()
}
self.finish(status: status)
}
}
nonisolated func locationManager(_ manager: CLLocationManager, didUpdateLocations locations: [CLLocation]) {
let status = manager.authorizationStatus
Task { @MainActor in
self.finish(status: status)
}
}
}
enum AppleScriptPermission {
private static let logger = Logger(subsystem: "bot.molt", category: "AppleScriptPermission")
/// Sends a benign AppleScript to Terminal to verify Automation permission.
@MainActor
static func isAuthorized() -> Bool {
let script = """
tell application "Terminal"
return "moltbot-ok"
end tell
"""
var error: NSDictionary?
let appleScript = NSAppleScript(source: script)
let result = appleScript?.executeAndReturnError(&error)
if let error, let code = error["NSAppleScriptErrorNumber"] as? Int {
if code == -1743 { // errAEEventWouldRequireUserConsent
Self.logger.debug("AppleScript permission denied (-1743)")
return false
}
Self.logger.debug("AppleScript check failed with code \(code)")
}
return result != nil
}
/// Triggers the TCC prompt and opens System Settings Privacy & Security Automation.
@MainActor
static func requestAuthorization() async {
_ = self.isAuthorized() // first attempt triggers the dialog if not granted
// Open the Automation pane to help the user if the prompt was dismissed.
let urlStrings = [
"x-apple.systempreferences:com.apple.preference.security?Privacy_Automation",
"x-apple.systempreferences:com.apple.preference.security",
]
for candidate in urlStrings {
if let url = URL(string: candidate), NSWorkspace.shared.open(url) {
break
}
}
}
}
@MainActor
@Observable
final class PermissionMonitor {
static let shared = PermissionMonitor()
private(set) var status: [Capability: Bool] = [:]
private var monitorTimer: Timer?
private var isChecking = false
private var registrations = 0
private var lastCheck: Date?
private let minimumCheckInterval: TimeInterval = 0.5
func register() {
self.registrations += 1
if self.registrations == 1 {
self.startMonitoring()
}
}
func unregister() {
guard self.registrations > 0 else { return }
self.registrations -= 1
if self.registrations == 0 {
self.stopMonitoring()
}
}
func refreshNow() async {
await self.checkStatus(force: true)
}
private func startMonitoring() {
Task { await self.checkStatus(force: true) }
if ProcessInfo.processInfo.isRunningTests {
return
}
self.monitorTimer = Timer.scheduledTimer(withTimeInterval: 1.0, repeats: true) { [weak self] _ in
guard let self else { return }
Task { @MainActor in
await self.checkStatus(force: false)
}
}
}
private func stopMonitoring() {
self.monitorTimer?.invalidate()
self.monitorTimer = nil
self.lastCheck = nil
}
private func checkStatus(force: Bool) async {
if self.isChecking { return }
let now = Date()
if !force, let lastCheck, now.timeIntervalSince(lastCheck) < self.minimumCheckInterval {
return
}
self.isChecking = true
let latest = await PermissionManager.status()
if latest != self.status {
self.status = latest
}
self.lastCheck = Date()
self.isChecking = false
}
}
enum ScreenRecordingProbe {
static func isAuthorized() -> Bool {
if #available(macOS 10.15, *) {
return CGPreflightScreenCaptureAccess()
}
return true
}
@MainActor
static func requestAuthorization() async {
if #available(macOS 10.15, *) {
_ = CGRequestScreenCaptureAccess()
}
}
}

View File

@@ -0,0 +1,418 @@
import Foundation
import OSLog
#if canImport(Darwin)
import Darwin
#endif
actor PortGuardian {
static let shared = PortGuardian()
struct Record: Codable {
let port: Int
let pid: Int32
let command: String
let mode: String
let timestamp: TimeInterval
}
struct Descriptor: Sendable {
let pid: Int32
let command: String
let executablePath: String?
}
private var records: [Record] = []
private let logger = Logger(subsystem: "bot.molt", category: "portguard")
private nonisolated static let appSupportDir: URL = {
let base = FileManager().urls(for: .applicationSupportDirectory, in: .userDomainMask).first!
return base.appendingPathComponent("Moltbot", isDirectory: true)
}()
private nonisolated static var recordPath: URL {
self.appSupportDir.appendingPathComponent("port-guard.json", isDirectory: false)
}
init() {
self.records = Self.loadRecords(from: Self.recordPath)
}
func sweep(mode: AppState.ConnectionMode) async {
self.logger.info("port sweep starting (mode=\(mode.rawValue, privacy: .public))")
guard mode != .unconfigured else {
self.logger.info("port sweep skipped (mode=unconfigured)")
return
}
let ports = [GatewayEnvironment.gatewayPort()]
for port in ports {
let listeners = await self.listeners(on: port)
guard !listeners.isEmpty else { continue }
for listener in listeners {
if self.isExpected(listener, port: port, mode: mode) {
let message = """
port \(port) already served by expected \(listener.command)
(pid \(listener.pid)) — keeping
"""
self.logger.info("\(message, privacy: .public)")
continue
}
let killed = await self.kill(listener.pid)
if killed {
let message = """
port \(port) was held by \(listener.command)
(pid \(listener.pid)); terminated
"""
self.logger.error("\(message, privacy: .public)")
} else {
self.logger.error("failed to terminate pid \(listener.pid) on port \(port, privacy: .public)")
}
}
}
self.logger.info("port sweep done")
}
func record(port: Int, pid: Int32, command: String, mode: AppState.ConnectionMode) async {
try? FileManager().createDirectory(at: Self.appSupportDir, withIntermediateDirectories: true)
self.records.removeAll { $0.pid == pid }
self.records.append(
Record(
port: port,
pid: pid,
command: command,
mode: mode.rawValue,
timestamp: Date().timeIntervalSince1970))
self.save()
}
func removeRecord(pid: Int32) {
let before = self.records.count
self.records.removeAll { $0.pid == pid }
if self.records.count != before {
self.save()
}
}
struct PortReport: Identifiable {
enum Status {
case ok(String)
case missing(String)
case interference(String, offenders: [ReportListener])
}
let port: Int
let expected: String
let status: Status
let listeners: [ReportListener]
var id: Int { self.port }
var offenders: [ReportListener] {
if case let .interference(_, offenders) = self.status { return offenders }
return []
}
var summary: String {
switch self.status {
case let .ok(text): text
case let .missing(text): text
case let .interference(text, _): text
}
}
}
func describe(port: Int) async -> Descriptor? {
guard let listener = await self.listeners(on: port).first else { return nil }
let path = Self.executablePath(for: listener.pid)
return Descriptor(pid: listener.pid, command: listener.command, executablePath: path)
}
// MARK: - Internals
private struct Listener {
let pid: Int32
let command: String
let fullCommand: String
let user: String?
}
struct ReportListener: Identifiable {
let pid: Int32
let command: String
let fullCommand: String
let user: String?
let expected: Bool
var id: Int32 { self.pid }
}
func diagnose(mode: AppState.ConnectionMode) async -> [PortReport] {
if mode == .unconfigured {
return []
}
let ports = [GatewayEnvironment.gatewayPort()]
var reports: [PortReport] = []
for port in ports {
let listeners = await self.listeners(on: port)
let tunnelHealthy = await self.probeGatewayHealthIfNeeded(
port: port,
mode: mode,
listeners: listeners)
reports.append(Self.buildReport(
port: port,
listeners: listeners,
mode: mode,
tunnelHealthy: tunnelHealthy))
}
return reports
}
func probeGatewayHealth(port: Int, timeout: TimeInterval = 2.0) async -> Bool {
let url = URL(string: "http://127.0.0.1:\(port)/")!
let config = URLSessionConfiguration.ephemeral
config.timeoutIntervalForRequest = timeout
config.timeoutIntervalForResource = timeout
let session = URLSession(configuration: config)
var request = URLRequest(url: url)
request.cachePolicy = .reloadIgnoringLocalCacheData
request.timeoutInterval = timeout
do {
let (_, response) = try await session.data(for: request)
return response is HTTPURLResponse
} catch {
return false
}
}
func isListening(port: Int, pid: Int32? = nil) async -> Bool {
let listeners = await self.listeners(on: port)
if let pid {
return listeners.contains(where: { $0.pid == pid })
}
return !listeners.isEmpty
}
private func listeners(on port: Int) async -> [Listener] {
let res = await ShellExecutor.run(
command: ["lsof", "-nP", "-iTCP:\(port)", "-sTCP:LISTEN", "-Fpcn"],
cwd: nil,
env: nil,
timeout: 5)
guard res.ok, let data = res.payload, !data.isEmpty else { return [] }
let text = String(data: data, encoding: .utf8) ?? ""
return Self.parseListeners(from: text)
}
private static func readFullCommand(pid: Int32) -> String? {
let proc = Process()
proc.executableURL = URL(fileURLWithPath: "/bin/ps")
proc.arguments = ["-p", "\(pid)", "-o", "command="]
let pipe = Pipe()
proc.standardOutput = pipe
proc.standardError = Pipe()
do {
let data = try proc.runAndReadToEnd(from: pipe)
guard !data.isEmpty else { return nil }
return String(data: data, encoding: .utf8)?
.trimmingCharacters(in: .whitespacesAndNewlines)
} catch {
return nil
}
}
private static func parseListeners(from text: String) -> [Listener] {
var listeners: [Listener] = []
var currentPid: Int32?
var currentCmd: String?
var currentUser: String?
func flush() {
if let pid = currentPid, let cmd = currentCmd {
let full = Self.readFullCommand(pid: pid) ?? cmd
listeners.append(Listener(pid: pid, command: cmd, fullCommand: full, user: currentUser))
}
currentPid = nil
currentCmd = nil
currentUser = nil
}
for line in text.split(separator: "\n") {
guard let prefix = line.first else { continue }
let value = String(line.dropFirst())
switch prefix {
case "p":
flush()
currentPid = Int32(value) ?? 0
case "c":
currentCmd = value
case "u":
currentUser = value
default:
continue
}
}
flush()
return listeners
}
private static func buildReport(
port: Int,
listeners: [Listener],
mode: AppState.ConnectionMode,
tunnelHealthy: Bool?) -> PortReport
{
let expectedDesc: String
let okPredicate: (Listener) -> Bool
let expectedCommands = ["node", "moltbot", "tsx", "pnpm", "bun"]
switch mode {
case .remote:
expectedDesc = "SSH tunnel to remote gateway"
okPredicate = { $0.command.lowercased().contains("ssh") }
case .local:
expectedDesc = "Gateway websocket (node/tsx)"
okPredicate = { listener in
let c = listener.command.lowercased()
return expectedCommands.contains { c.contains($0) }
}
case .unconfigured:
expectedDesc = "Gateway not configured"
okPredicate = { _ in false }
}
if listeners.isEmpty {
let text = "Nothing is listening on \(port) (\(expectedDesc))."
return .init(port: port, expected: expectedDesc, status: .missing(text), listeners: [])
}
let tunnelUnhealthy =
mode == .remote && port == GatewayEnvironment.gatewayPort() && tunnelHealthy == false
let reportListeners = listeners.map { listener in
var expected = okPredicate(listener)
if tunnelUnhealthy, expected { expected = false }
return ReportListener(
pid: listener.pid,
command: listener.command,
fullCommand: listener.fullCommand,
user: listener.user,
expected: expected)
}
let offenders = reportListeners.filter { !$0.expected }
if tunnelUnhealthy {
let list = listeners.map { "\($0.command) (\($0.pid))" }.joined(separator: ", ")
let reason = "Port \(port) is served by \(list), but the SSH tunnel is unhealthy."
return .init(
port: port,
expected: expectedDesc,
status: .interference(reason, offenders: offenders),
listeners: reportListeners)
}
if offenders.isEmpty {
let list = listeners.map { "\($0.command) (\($0.pid))" }.joined(separator: ", ")
let okText = "Port \(port) is served by \(list)."
return .init(
port: port,
expected: expectedDesc,
status: .ok(okText),
listeners: reportListeners)
}
let list = offenders.map { "\($0.command) (\($0.pid))" }.joined(separator: ", ")
let reason = "Port \(port) is held by \(list), expected \(expectedDesc)."
return .init(
port: port,
expected: expectedDesc,
status: .interference(reason, offenders: offenders),
listeners: reportListeners)
}
private static func executablePath(for pid: Int32) -> String? {
#if canImport(Darwin)
var buffer = [CChar](repeating: 0, count: Int(PATH_MAX))
let length = proc_pidpath(pid, &buffer, UInt32(buffer.count))
guard length > 0 else { return nil }
// Drop trailing null and decode as UTF-8.
let trimmed = buffer.prefix { $0 != 0 }
let bytes = trimmed.map { UInt8(bitPattern: $0) }
return String(bytes: bytes, encoding: .utf8)
#else
return nil
#endif
}
private func kill(_ pid: Int32) async -> Bool {
let term = await ShellExecutor.run(command: ["kill", "-TERM", "\(pid)"], cwd: nil, env: nil, timeout: 2)
if term.ok { return true }
let sigkill = await ShellExecutor.run(command: ["kill", "-KILL", "\(pid)"], cwd: nil, env: nil, timeout: 2)
return sigkill.ok
}
private func isExpected(_ listener: Listener, port: Int, mode: AppState.ConnectionMode) -> Bool {
let cmd = listener.command.lowercased()
let full = listener.fullCommand.lowercased()
switch mode {
case .remote:
// Remote mode expects an SSH tunnel for the gateway WebSocket port.
if port == GatewayEnvironment.gatewayPort() { return cmd.contains("ssh") }
return false
case .local:
// The gateway daemon may listen as `moltbot` or as its runtime (`node`, `bun`, etc).
if full.contains("gateway-daemon") { return true }
// If args are unavailable, treat a moltbot listener as expected.
if cmd.contains("moltbot"), full == cmd { return true }
return false
case .unconfigured:
return false
}
}
private func probeGatewayHealthIfNeeded(
port: Int,
mode: AppState.ConnectionMode,
listeners: [Listener]) async -> Bool?
{
guard mode == .remote, port == GatewayEnvironment.gatewayPort(), !listeners.isEmpty else { return nil }
let hasSsh = listeners.contains { $0.command.lowercased().contains("ssh") }
guard hasSsh else { return nil }
return await self.probeGatewayHealth(port: port)
}
private static func loadRecords(from url: URL) -> [Record] {
guard let data = try? Data(contentsOf: url),
let decoded = try? JSONDecoder().decode([Record].self, from: data)
else { return [] }
return decoded
}
private func save() {
guard let data = try? JSONEncoder().encode(self.records) else { return }
try? data.write(to: Self.recordPath, options: [.atomic])
}
}
#if DEBUG
extension PortGuardian {
static func _testParseListeners(_ text: String) -> [(
pid: Int32,
command: String,
fullCommand: String,
user: String?)]
{
self.parseListeners(from: text).map { ($0.pid, $0.command, $0.fullCommand, $0.user) }
}
static func _testBuildReport(
port: Int,
mode: AppState.ConnectionMode,
listeners: [(pid: Int32, command: String, fullCommand: String, user: String?)]) -> PortReport
{
let mapped = listeners.map { Listener(
pid: $0.pid,
command: $0.command,
fullCommand: $0.fullCommand,
user: $0.user) }
return Self.buildReport(port: port, listeners: mapped, mode: mode, tunnelHealthy: nil)
}
}
#endif

View File

@@ -0,0 +1,158 @@
import Cocoa
import Darwin
import Foundation
import OSLog
@MainActor
final class PresenceReporter {
static let shared = PresenceReporter()
private let logger = Logger(subsystem: "bot.molt", category: "presence")
private var task: Task<Void, Never>?
private let interval: TimeInterval = 180 // a few minutes
private let instanceId: String = InstanceIdentity.instanceId
func start() {
guard self.task == nil else { return }
self.task = Task.detached { [weak self] in
guard let self else { return }
await self.push(reason: "launch")
while !Task.isCancelled {
try? await Task.sleep(nanoseconds: UInt64(self.interval * 1_000_000_000))
await self.push(reason: "periodic")
}
}
}
func stop() {
self.task?.cancel()
self.task = nil
}
@Sendable
private func push(reason: String) async {
let mode = await MainActor.run { AppStateStore.shared.connectionMode.rawValue }
let host = InstanceIdentity.displayName
let ip = Self.primaryIPv4Address() ?? "ip-unknown"
let version = Self.appVersionString()
let platform = Self.platformString()
let lastInput = Self.lastInputSeconds()
let text = Self.composePresenceSummary(mode: mode, reason: reason)
var params: [String: AnyHashable] = [
"instanceId": AnyHashable(self.instanceId),
"host": AnyHashable(host),
"ip": AnyHashable(ip),
"mode": AnyHashable(mode),
"version": AnyHashable(version),
"platform": AnyHashable(platform),
"deviceFamily": AnyHashable("Mac"),
"reason": AnyHashable(reason),
]
if let model = InstanceIdentity.modelIdentifier { params["modelIdentifier"] = AnyHashable(model) }
if let lastInput { params["lastInputSeconds"] = AnyHashable(lastInput) }
do {
try await ControlChannel.shared.sendSystemEvent(text, params: params)
} catch {
self.logger.error("presence send failed: \(error.localizedDescription, privacy: .public)")
}
}
/// Fire an immediate presence beacon (e.g., right after connecting).
func sendImmediate(reason: String = "connect") {
Task { await self.push(reason: reason) }
}
private static func composePresenceSummary(mode: String, reason: String) -> String {
let host = InstanceIdentity.displayName
let ip = Self.primaryIPv4Address() ?? "ip-unknown"
let version = Self.appVersionString()
let lastInput = Self.lastInputSeconds()
let lastLabel = lastInput.map { "last input \($0)s ago" } ?? "last input unknown"
return "Node: \(host) (\(ip)) · app \(version) · \(lastLabel) · mode \(mode) · reason \(reason)"
}
private static func appVersionString() -> String {
let version = Bundle.main.object(forInfoDictionaryKey: "CFBundleShortVersionString") as? String ?? "dev"
if let build = Bundle.main.object(forInfoDictionaryKey: "CFBundleVersion") as? String {
let trimmed = build.trimmingCharacters(in: .whitespacesAndNewlines)
if !trimmed.isEmpty, trimmed != version {
return "\(version) (\(trimmed))"
}
}
return version
}
private static func platformString() -> String {
let v = ProcessInfo.processInfo.operatingSystemVersion
return "macos \(v.majorVersion).\(v.minorVersion).\(v.patchVersion)"
}
private static func lastInputSeconds() -> Int? {
let anyEvent = CGEventType(rawValue: UInt32.max) ?? .null
let seconds = CGEventSource.secondsSinceLastEventType(.combinedSessionState, eventType: anyEvent)
if seconds.isNaN || seconds.isInfinite || seconds < 0 { return nil }
return Int(seconds.rounded())
}
private static func primaryIPv4Address() -> String? {
var addrList: UnsafeMutablePointer<ifaddrs>?
guard getifaddrs(&addrList) == 0, let first = addrList else { return nil }
defer { freeifaddrs(addrList) }
var fallback: String?
var en0: String?
for ptr in sequence(first: first, next: { $0.pointee.ifa_next }) {
let flags = Int32(ptr.pointee.ifa_flags)
let isUp = (flags & IFF_UP) != 0
let isLoopback = (flags & IFF_LOOPBACK) != 0
let name = String(cString: ptr.pointee.ifa_name)
let family = ptr.pointee.ifa_addr.pointee.sa_family
if !isUp || isLoopback || family != UInt8(AF_INET) { continue }
var addr = ptr.pointee.ifa_addr.pointee
var buffer = [CChar](repeating: 0, count: Int(NI_MAXHOST))
let result = getnameinfo(
&addr,
socklen_t(ptr.pointee.ifa_addr.pointee.sa_len),
&buffer,
socklen_t(buffer.count),
nil,
0,
NI_NUMERICHOST)
guard result == 0 else { continue }
let len = buffer.prefix { $0 != 0 }
let bytes = len.map { UInt8(bitPattern: $0) }
guard let ip = String(bytes: bytes, encoding: .utf8) else { continue }
if name == "en0" { en0 = ip; break }
if fallback == nil { fallback = ip }
}
return en0 ?? fallback
}
}
#if DEBUG
extension PresenceReporter {
static func _testComposePresenceSummary(mode: String, reason: String) -> String {
self.composePresenceSummary(mode: mode, reason: reason)
}
static func _testAppVersionString() -> String {
self.appVersionString()
}
static func _testPlatformString() -> String {
self.platformString()
}
static func _testLastInputSeconds() -> Int? {
self.lastInputSeconds()
}
static func _testPrimaryIPv4Address() -> String? {
self.primaryIPv4Address()
}
}
#endif

View File

@@ -0,0 +1,317 @@
import Foundation
import Network
import OSLog
#if canImport(Darwin)
import Darwin
#endif
/// Port forwarding tunnel for remote mode.
///
/// Uses `ssh -N -L` to forward the remote gateway ports to localhost.
final class RemotePortTunnel {
private static let logger = Logger(subsystem: "bot.molt", category: "remote.tunnel")
let process: Process
let localPort: UInt16?
private let stderrHandle: FileHandle?
private init(process: Process, localPort: UInt16?, stderrHandle: FileHandle?) {
self.process = process
self.localPort = localPort
self.stderrHandle = stderrHandle
}
deinit {
Self.cleanupStderr(self.stderrHandle)
let pid = self.process.processIdentifier
self.process.terminate()
Task { await PortGuardian.shared.removeRecord(pid: pid) }
}
func terminate() {
Self.cleanupStderr(self.stderrHandle)
let pid = self.process.processIdentifier
if self.process.isRunning {
self.process.terminate()
self.process.waitUntilExit()
}
Task { await PortGuardian.shared.removeRecord(pid: pid) }
}
static func create(
remotePort: Int,
preferredLocalPort: UInt16? = nil,
allowRemoteUrlOverride: Bool = true,
allowRandomLocalPort: Bool = true) async throws -> RemotePortTunnel
{
let settings = CommandResolver.connectionSettings()
guard settings.mode == .remote, let parsed = CommandResolver.parseSSHTarget(settings.target) else {
throw NSError(
domain: "RemotePortTunnel",
code: 3,
userInfo: [NSLocalizedDescriptionKey: "Remote mode is not configured"])
}
let localPort = try await Self.findPort(
preferred: preferredLocalPort,
allowRandom: allowRandomLocalPort)
let sshHost = parsed.host.trimmingCharacters(in: .whitespacesAndNewlines)
let remotePortOverride =
allowRemoteUrlOverride && remotePort == GatewayEnvironment.gatewayPort()
? Self.resolveRemotePortOverride(for: sshHost)
: nil
let resolvedRemotePort = remotePortOverride ?? remotePort
if let override = remotePortOverride {
Self.logger.info(
"ssh tunnel remote port override " +
"host=\(sshHost, privacy: .public) port=\(override, privacy: .public)")
} else {
Self.logger.debug(
"ssh tunnel using default remote port " +
"host=\(sshHost, privacy: .public) port=\(remotePort, privacy: .public)")
}
let options: [String] = [
"-o", "BatchMode=yes",
"-o", "ExitOnForwardFailure=yes",
"-o", "StrictHostKeyChecking=accept-new",
"-o", "UpdateHostKeys=yes",
"-o", "ServerAliveInterval=15",
"-o", "ServerAliveCountMax=3",
"-o", "TCPKeepAlive=yes",
"-N",
"-L", "\(localPort):127.0.0.1:\(resolvedRemotePort)",
]
let identity = settings.identity.trimmingCharacters(in: .whitespacesAndNewlines)
let args = CommandResolver.sshArguments(
target: parsed,
identity: identity,
options: options)
let process = Process()
process.executableURL = URL(fileURLWithPath: "/usr/bin/ssh")
process.arguments = args
let pipe = Pipe()
process.standardError = pipe
let stderrHandle = pipe.fileHandleForReading
// Consume stderr so ssh cannot block if it logs.
stderrHandle.readabilityHandler = { handle in
let data = handle.readSafely(upToCount: 64 * 1024)
guard !data.isEmpty else {
// EOF (or read failure): stop monitoring to avoid spinning on a closed pipe.
Self.cleanupStderr(handle)
return
}
guard let line = String(data: data, encoding: .utf8)?
.trimmingCharacters(in: .whitespacesAndNewlines),
!line.isEmpty
else { return }
Self.logger.error("ssh tunnel stderr: \(line, privacy: .public)")
}
process.terminationHandler = { _ in
Self.cleanupStderr(stderrHandle)
}
try process.run()
// If ssh exits immediately (e.g. local port already in use), surface stderr and ensure we stop monitoring.
try? await Task.sleep(nanoseconds: 150_000_000) // 150ms
if !process.isRunning {
let stderr = Self.drainStderr(stderrHandle)
let msg = stderr.isEmpty ? "ssh tunnel exited immediately" : "ssh tunnel failed: \(stderr)"
throw NSError(domain: "RemotePortTunnel", code: 4, userInfo: [NSLocalizedDescriptionKey: msg])
}
// Track tunnel so we can clean up stale listeners on restart.
Task {
await PortGuardian.shared.record(
port: Int(localPort),
pid: process.processIdentifier,
command: process.executableURL?.path ?? "ssh",
mode: CommandResolver.connectionSettings().mode)
}
return RemotePortTunnel(process: process, localPort: localPort, stderrHandle: stderrHandle)
}
private static func resolveRemotePortOverride(for sshHost: String) -> Int? {
let root = MoltbotConfigFile.loadDict()
guard let gateway = root["gateway"] as? [String: Any],
let remote = gateway["remote"] as? [String: Any],
let urlRaw = remote["url"] as? String
else {
return nil
}
let trimmed = urlRaw.trimmingCharacters(in: .whitespacesAndNewlines)
guard !trimmed.isEmpty, let url = URL(string: trimmed), let port = url.port else {
return nil
}
guard let host = url.host?.trimmingCharacters(in: .whitespacesAndNewlines),
!host.isEmpty
else {
return nil
}
let sshKey = Self.hostKey(sshHost)
let urlKey = Self.hostKey(host)
guard !sshKey.isEmpty, !urlKey.isEmpty else { return nil }
guard sshKey == urlKey else {
Self.logger.debug(
"remote url host mismatch sshHost=\(sshHost, privacy: .public) urlHost=\(host, privacy: .public)")
return nil
}
return port
}
private static func hostKey(_ host: String) -> String {
let trimmed = host.trimmingCharacters(in: .whitespacesAndNewlines).lowercased()
guard !trimmed.isEmpty else { return "" }
if trimmed.contains(":") { return trimmed }
let digits = CharacterSet(charactersIn: "0123456789.")
if trimmed.rangeOfCharacter(from: digits.inverted) == nil {
return trimmed
}
return trimmed.split(separator: ".").first.map(String.init) ?? trimmed
}
private static func findPort(preferred: UInt16?, allowRandom: Bool) async throws -> UInt16 {
if let preferred, self.portIsFree(preferred) { return preferred }
if let preferred, !allowRandom {
throw NSError(
domain: "RemotePortTunnel",
code: 5,
userInfo: [
NSLocalizedDescriptionKey: "Local port \(preferred) is unavailable",
])
}
return try await withCheckedThrowingContinuation { cont in
let queue = DispatchQueue(label: "bot.molt.remote.tunnel.port", qos: .utility)
do {
let listener = try NWListener(using: .tcp, on: .any)
listener.newConnectionHandler = { connection in connection.cancel() }
listener.stateUpdateHandler = { state in
switch state {
case .ready:
if let port = listener.port?.rawValue {
listener.stateUpdateHandler = nil
listener.cancel()
cont.resume(returning: port)
}
case let .failed(error):
listener.stateUpdateHandler = nil
listener.cancel()
cont.resume(throwing: error)
default:
break
}
}
listener.start(queue: queue)
} catch {
cont.resume(throwing: error)
}
}
}
private static func portIsFree(_ port: UInt16) -> Bool {
#if canImport(Darwin)
// NWListener can succeed even when only one address family is held. Mirror what ssh needs by checking
// both 127.0.0.1 and ::1 for availability.
return self.canBindIPv4(port) && self.canBindIPv6(port)
#else
do {
let listener = try NWListener(using: .tcp, on: NWEndpoint.Port(rawValue: port)!)
listener.cancel()
return true
} catch {
return false
}
#endif
}
#if canImport(Darwin)
private static func canBindIPv4(_ port: UInt16) -> Bool {
let fd = socket(AF_INET, SOCK_STREAM, 0)
guard fd >= 0 else { return false }
defer { _ = Darwin.close(fd) }
var one: Int32 = 1
_ = setsockopt(fd, SOL_SOCKET, SO_REUSEADDR, &one, socklen_t(MemoryLayout.size(ofValue: one)))
var addr = sockaddr_in()
addr.sin_len = UInt8(MemoryLayout<sockaddr_in>.size)
addr.sin_family = sa_family_t(AF_INET)
addr.sin_port = port.bigEndian
addr.sin_addr = in_addr(s_addr: inet_addr("127.0.0.1"))
let result = withUnsafePointer(to: &addr) { ptr in
ptr.withMemoryRebound(to: sockaddr.self, capacity: 1) { sa in
Darwin.bind(fd, sa, socklen_t(MemoryLayout<sockaddr_in>.size))
}
}
return result == 0
}
private static func canBindIPv6(_ port: UInt16) -> Bool {
let fd = socket(AF_INET6, SOCK_STREAM, 0)
guard fd >= 0 else { return false }
defer { _ = Darwin.close(fd) }
var one: Int32 = 1
_ = setsockopt(fd, SOL_SOCKET, SO_REUSEADDR, &one, socklen_t(MemoryLayout.size(ofValue: one)))
var addr = sockaddr_in6()
addr.sin6_len = UInt8(MemoryLayout<sockaddr_in6>.size)
addr.sin6_family = sa_family_t(AF_INET6)
addr.sin6_port = port.bigEndian
var loopback = in6_addr()
_ = withUnsafeMutablePointer(to: &loopback) { ptr in
inet_pton(AF_INET6, "::1", ptr)
}
addr.sin6_addr = loopback
let result = withUnsafePointer(to: &addr) { ptr in
ptr.withMemoryRebound(to: sockaddr.self, capacity: 1) { sa in
Darwin.bind(fd, sa, socklen_t(MemoryLayout<sockaddr_in6>.size))
}
}
return result == 0
}
#endif
private static func cleanupStderr(_ handle: FileHandle?) {
guard let handle else { return }
Self.cleanupStderr(handle)
}
private static func cleanupStderr(_ handle: FileHandle) {
if handle.readabilityHandler != nil {
handle.readabilityHandler = nil
}
try? handle.close()
}
private static func drainStderr(_ handle: FileHandle) -> String {
handle.readabilityHandler = nil
defer { try? handle.close() }
do {
let data = try handle.readToEnd() ?? Data()
return String(data: data, encoding: .utf8)?
.trimmingCharacters(in: .whitespacesAndNewlines) ?? ""
} catch {
self.logger.debug("Failed to drain ssh stderr: \(error, privacy: .public)")
return ""
}
}
#if SWIFT_PACKAGE
static func _testPortIsFree(_ port: UInt16) -> Bool {
self.portIsFree(port)
}
static func _testDrainStderr(_ handle: FileHandle) -> String {
self.drainStderr(handle)
}
#endif
}

View File

@@ -0,0 +1,122 @@
import Foundation
import OSLog
/// Manages the SSH tunnel that forwards the remote gateway/control port to localhost.
actor RemoteTunnelManager {
static let shared = RemoteTunnelManager()
private let logger = Logger(subsystem: "bot.molt", category: "remote-tunnel")
private var controlTunnel: RemotePortTunnel?
private var restartInFlight = false
private var lastRestartAt: Date?
private let restartBackoffSeconds: TimeInterval = 2.0
func controlTunnelPortIfRunning() async -> UInt16? {
if self.restartInFlight {
self.logger.info("control tunnel restart in flight; skipping reuse check")
return nil
}
if let tunnel = self.controlTunnel,
tunnel.process.isRunning,
let local = tunnel.localPort
{
let pid = tunnel.process.processIdentifier
if await PortGuardian.shared.isListening(port: Int(local), pid: pid) {
self.logger.info("reusing active SSH tunnel localPort=\(local, privacy: .public)")
return local
}
self.logger.error(
"active SSH tunnel on port \(local, privacy: .public) is not listening; restarting")
await self.beginRestart()
tunnel.terminate()
self.controlTunnel = nil
}
// If a previous Moltbot run already has an SSH listener on the expected port (common after restarts),
// reuse it instead of spawning new ssh processes that immediately fail with "Address already in use".
let desiredPort = UInt16(GatewayEnvironment.gatewayPort())
if let desc = await PortGuardian.shared.describe(port: Int(desiredPort)),
self.isSshProcess(desc)
{
self.logger.info(
"reusing existing SSH tunnel listener " +
"localPort=\(desiredPort, privacy: .public) " +
"pid=\(desc.pid, privacy: .public)")
return desiredPort
}
return nil
}
/// Ensure an SSH tunnel is running for the gateway control port.
/// Returns the local forwarded port (usually the configured gateway port).
func ensureControlTunnel() async throws -> UInt16 {
let settings = CommandResolver.connectionSettings()
guard settings.mode == .remote else {
throw NSError(
domain: "RemoteTunnel",
code: 1,
userInfo: [NSLocalizedDescriptionKey: "Remote mode is not enabled"])
}
let identitySet = !settings.identity.trimmingCharacters(in: .whitespacesAndNewlines).isEmpty
self.logger.info(
"ensure SSH tunnel target=\(settings.target, privacy: .public) " +
"identitySet=\(identitySet, privacy: .public)")
if let local = await self.controlTunnelPortIfRunning() { return local }
await self.waitForRestartBackoffIfNeeded()
let desiredPort = UInt16(GatewayEnvironment.gatewayPort())
let tunnel = try await RemotePortTunnel.create(
remotePort: GatewayEnvironment.gatewayPort(),
preferredLocalPort: desiredPort,
allowRandomLocalPort: false)
self.controlTunnel = tunnel
self.endRestart()
let resolvedPort = tunnel.localPort ?? desiredPort
self.logger.info("ssh tunnel ready localPort=\(resolvedPort, privacy: .public)")
return tunnel.localPort ?? desiredPort
}
func stopAll() {
self.controlTunnel?.terminate()
self.controlTunnel = nil
}
private func isSshProcess(_ desc: PortGuardian.Descriptor) -> Bool {
let cmd = desc.command.lowercased()
if cmd.contains("ssh") { return true }
if let path = desc.executablePath?.lowercased(), path.contains("/ssh") { return true }
return false
}
private func beginRestart() async {
guard !self.restartInFlight else { return }
self.restartInFlight = true
self.lastRestartAt = Date()
self.logger.info("control tunnel restart started")
Task { [weak self] in
guard let self else { return }
try? await Task.sleep(nanoseconds: UInt64(self.restartBackoffSeconds * 1_000_000_000))
await self.endRestart()
}
}
private func endRestart() {
if self.restartInFlight {
self.restartInFlight = false
self.logger.info("control tunnel restart finished")
}
}
private func waitForRestartBackoffIfNeeded() async {
guard let last = self.lastRestartAt else { return }
let elapsed = Date().timeIntervalSince(last)
let remaining = self.restartBackoffSeconds - elapsed
guard remaining > 0 else { return }
self.logger.info(
"control tunnel restart backoff \(remaining, privacy: .public)s")
try? await Task.sleep(nanoseconds: UInt64(remaining * 1_000_000_000))
}
// Keep tunnel reuse lightweight; restart only when the listener disappears.
}

View File

@@ -0,0 +1,79 @@
<?xml version="1.0" encoding="UTF-8"?>
<!DOCTYPE plist PUBLIC "-//Apple//DTD PLIST 1.0//EN" "http://www.apple.com/DTDs/PropertyList-1.0.dtd">
<plist version="1.0">
<dict>
<key>CFBundleDevelopmentRegion</key>
<string>en</string>
<key>CFBundleExecutable</key>
<string>Moltbot</string>
<key>CFBundleIdentifier</key>
<string>bot.molt.mac</string>
<key>CFBundleInfoDictionaryVersion</key>
<string>6.0</string>
<key>CFBundleName</key>
<string>Moltbot</string>
<key>CFBundlePackageType</key>
<string>APPL</string>
<key>CFBundleShortVersionString</key>
<string>2026.1.26</string>
<key>CFBundleVersion</key>
<string>202601260</string>
<key>CFBundleIconFile</key>
<string>Moltbot</string>
<key>CFBundleURLTypes</key>
<array>
<dict>
<key>CFBundleURLName</key>
<string>bot.molt.mac.deeplink</string>
<key>CFBundleURLSchemes</key>
<array>
<string>moltbot</string>
</array>
</dict>
</array>
<key>LSMinimumSystemVersion</key>
<string>15.0</string>
<key>LSUIElement</key>
<true/>
<key>MoltbotBuildTimestamp</key>
<string></string>
<key>MoltbotGitCommit</key>
<string></string>
<key>NSUserNotificationUsageDescription</key>
<string>Moltbot needs notification permission to show alerts for agent actions.</string>
<key>NSScreenCaptureDescription</key>
<string>Moltbot captures the screen when the agent needs screenshots for context.</string>
<key>NSCameraUsageDescription</key>
<string>Moltbot can capture photos or short video clips when requested by the agent.</string>
<key>NSLocationUsageDescription</key>
<string>Moltbot can share your location when requested by the agent.</string>
<key>NSLocationWhenInUseUsageDescription</key>
<string>Moltbot can share your location when requested by the agent.</string>
<key>NSLocationAlwaysAndWhenInUseUsageDescription</key>
<string>Moltbot can share your location when requested by the agent.</string>
<key>NSMicrophoneUsageDescription</key>
<string>Moltbot needs the mic for Voice Wake tests and agent audio capture.</string>
<key>NSSpeechRecognitionUsageDescription</key>
<string>Moltbot uses speech recognition to detect your Voice Wake trigger phrase.</string>
<key>NSAppleEventsUsageDescription</key>
<string>Moltbot needs Automation (AppleScript) permission to drive Terminal and other apps for agent actions.</string>
<key>NSAppTransportSecurity</key>
<dict>
<key>NSAllowsArbitraryLoadsInWebContent</key>
<true/>
<key>NSExceptionDomains</key>
<dict>
<key>100.100.100.100</key>
<dict>
<key>NSExceptionAllowsInsecureHTTPLoads</key>
<true/>
<key>NSIncludesSubdomains</key>
<true/>
</dict>
</dict>
</dict>
</dict>
</plist>

View File

@@ -0,0 +1,167 @@
import Foundation
import OSLog
enum RuntimeKind: String {
case node
}
struct RuntimeVersion: Comparable, CustomStringConvertible {
let major: Int
let minor: Int
let patch: Int
var description: String { "\(self.major).\(self.minor).\(self.patch)" }
static func < (lhs: RuntimeVersion, rhs: RuntimeVersion) -> Bool {
if lhs.major != rhs.major { return lhs.major < rhs.major }
if lhs.minor != rhs.minor { return lhs.minor < rhs.minor }
return lhs.patch < rhs.patch
}
static func from(string: String) -> RuntimeVersion? {
// Accept optional leading "v" and ignore trailing metadata.
let pattern = #"(\d+)\.(\d+)\.(\d+)"#
guard let match = string.range(of: pattern, options: .regularExpression) else { return nil }
let versionString = String(string[match])
let parts = versionString.split(separator: ".")
guard parts.count == 3,
let major = Int(parts[0]),
let minor = Int(parts[1]),
let patch = Int(parts[2])
else { return nil }
return RuntimeVersion(major: major, minor: minor, patch: patch)
}
}
struct RuntimeResolution {
let kind: RuntimeKind
let path: String
let version: RuntimeVersion
}
enum RuntimeResolutionError: Error {
case notFound(searchPaths: [String])
case unsupported(
kind: RuntimeKind,
found: RuntimeVersion,
required: RuntimeVersion,
path: String,
searchPaths: [String])
case versionParse(kind: RuntimeKind, raw: String, path: String, searchPaths: [String])
}
enum RuntimeLocator {
private static let logger = Logger(subsystem: "bot.molt", category: "runtime")
private static let minNode = RuntimeVersion(major: 22, minor: 0, patch: 0)
static func resolve(
searchPaths: [String] = CommandResolver.preferredPaths()) -> Result<RuntimeResolution, RuntimeResolutionError>
{
let pathEnv = searchPaths.joined(separator: ":")
let runtime: RuntimeKind = .node
guard let binary = findExecutable(named: runtime.binaryName, searchPaths: searchPaths) else {
return .failure(.notFound(searchPaths: searchPaths))
}
guard let rawVersion = readVersion(of: binary, pathEnv: pathEnv) else {
return .failure(.versionParse(
kind: runtime,
raw: "(unreadable)",
path: binary,
searchPaths: searchPaths))
}
guard let parsed = RuntimeVersion.from(string: rawVersion) else {
return .failure(.versionParse(kind: runtime, raw: rawVersion, path: binary, searchPaths: searchPaths))
}
guard parsed >= self.minNode else {
return .failure(.unsupported(
kind: runtime,
found: parsed,
required: self.minNode,
path: binary,
searchPaths: searchPaths))
}
return .success(RuntimeResolution(kind: runtime, path: binary, version: parsed))
}
static func describeFailure(_ error: RuntimeResolutionError) -> String {
switch error {
case let .notFound(searchPaths):
[
"moltbot needs Node >=22.0.0 but found no runtime.",
"PATH searched: \(searchPaths.joined(separator: ":"))",
"Install Node: https://nodejs.org/en/download",
].joined(separator: "\n")
case let .unsupported(kind, found, required, path, searchPaths):
[
"Found \(kind.rawValue) \(found) at \(path) but need >= \(required).",
"PATH searched: \(searchPaths.joined(separator: ":"))",
"Upgrade Node and rerun moltbot.",
].joined(separator: "\n")
case let .versionParse(kind, raw, path, searchPaths):
[
"Could not parse \(kind.rawValue) version output \"\(raw)\" from \(path).",
"PATH searched: \(searchPaths.joined(separator: ":"))",
"Try reinstalling or pinning a supported version (Node >=22.0.0).",
].joined(separator: "\n")
}
}
// MARK: - Internals
private static func findExecutable(named name: String, searchPaths: [String]) -> String? {
let fm = FileManager()
for dir in searchPaths {
let candidate = (dir as NSString).appendingPathComponent(name)
if fm.isExecutableFile(atPath: candidate) {
return candidate
}
}
return nil
}
private static func readVersion(of binary: String, pathEnv: String) -> String? {
let start = Date()
let process = Process()
process.executableURL = URL(fileURLWithPath: binary)
process.arguments = ["--version"]
process.environment = ["PATH": pathEnv]
let pipe = Pipe()
process.standardOutput = pipe
process.standardError = pipe
do {
let data = try process.runAndReadToEnd(from: pipe)
let elapsedMs = Int(Date().timeIntervalSince(start) * 1000)
if elapsedMs > 500 {
self.logger.warning(
"""
runtime --version slow (\(elapsedMs, privacy: .public)ms) \
bin=\(binary, privacy: .public)
""")
} else {
self.logger.debug(
"""
runtime --version ok (\(elapsedMs, privacy: .public)ms) \
bin=\(binary, privacy: .public)
""")
}
return String(data: data, encoding: .utf8)?.trimmingCharacters(in: .whitespacesAndNewlines)
} catch {
let elapsedMs = Int(Date().timeIntervalSince(start) * 1000)
self.logger.error(
"""
runtime --version failed (\(elapsedMs, privacy: .public)ms) \
bin=\(binary, privacy: .public) \
err=\(error.localizedDescription, privacy: .public)
""")
return nil
}
}
}
extension RuntimeKind {
fileprivate var binaryName: String { "node" }
}

View File

@@ -0,0 +1,266 @@
import AVFoundation
import Foundation
import OSLog
@preconcurrency import ScreenCaptureKit
@MainActor
final class ScreenRecordService {
enum ScreenRecordError: LocalizedError {
case noDisplays
case invalidScreenIndex(Int)
case noFramesCaptured
case writeFailed(String)
var errorDescription: String? {
switch self {
case .noDisplays:
"No displays available for screen recording"
case let .invalidScreenIndex(idx):
"Invalid screen index \(idx)"
case .noFramesCaptured:
"No frames captured"
case let .writeFailed(msg):
msg
}
}
}
private let logger = Logger(subsystem: "bot.molt", category: "screenRecord")
func record(
screenIndex: Int?,
durationMs: Int?,
fps: Double?,
includeAudio: Bool?,
outPath: String?) async throws -> (path: String, hasAudio: Bool)
{
let durationMs = Self.clampDurationMs(durationMs)
let fps = Self.clampFps(fps)
let includeAudio = includeAudio ?? false
let outURL: URL = {
if let outPath, !outPath.trimmingCharacters(in: .whitespacesAndNewlines).isEmpty {
return URL(fileURLWithPath: outPath)
}
return FileManager().temporaryDirectory
.appendingPathComponent("moltbot-screen-record-\(UUID().uuidString).mp4")
}()
try? FileManager().removeItem(at: outURL)
let content = try await SCShareableContent.current
let displays = content.displays.sorted { $0.displayID < $1.displayID }
guard !displays.isEmpty else { throw ScreenRecordError.noDisplays }
let idx = screenIndex ?? 0
guard idx >= 0, idx < displays.count else { throw ScreenRecordError.invalidScreenIndex(idx) }
let display = displays[idx]
let filter = SCContentFilter(display: display, excludingWindows: [])
let config = SCStreamConfiguration()
config.width = display.width
config.height = display.height
config.queueDepth = 8
config.showsCursor = true
config.minimumFrameInterval = CMTime(value: 1, timescale: CMTimeScale(max(1, Int32(fps.rounded()))))
if includeAudio {
config.capturesAudio = true
}
let recorder = try StreamRecorder(
outputURL: outURL,
width: display.width,
height: display.height,
includeAudio: includeAudio,
logger: self.logger)
let stream = SCStream(filter: filter, configuration: config, delegate: recorder)
try stream.addStreamOutput(recorder, type: .screen, sampleHandlerQueue: recorder.queue)
if includeAudio {
try stream.addStreamOutput(recorder, type: .audio, sampleHandlerQueue: recorder.queue)
}
self.logger.info(
"screen record start idx=\(idx) durationMs=\(durationMs) fps=\(fps) out=\(outURL.path, privacy: .public)")
var started = false
do {
try await stream.startCapture()
started = true
try await Task.sleep(nanoseconds: UInt64(durationMs) * 1_000_000)
try await stream.stopCapture()
} catch {
if started { try? await stream.stopCapture() }
throw error
}
try await recorder.finish()
return (path: outURL.path, hasAudio: recorder.hasAudio)
}
private nonisolated static func clampDurationMs(_ ms: Int?) -> Int {
let v = ms ?? 10000
return min(60000, max(250, v))
}
private nonisolated static func clampFps(_ fps: Double?) -> Double {
let v = fps ?? 10
if !v.isFinite { return 10 }
return min(60, max(1, v))
}
}
private final class StreamRecorder: NSObject, SCStreamOutput, SCStreamDelegate, @unchecked Sendable {
let queue = DispatchQueue(label: "bot.molt.screenRecord.writer")
private let logger: Logger
private let writer: AVAssetWriter
private let input: AVAssetWriterInput
private let audioInput: AVAssetWriterInput?
let hasAudio: Bool
private var started = false
private var sawFrame = false
private var didFinish = false
private var pendingErrorMessage: String?
init(outputURL: URL, width: Int, height: Int, includeAudio: Bool, logger: Logger) throws {
self.logger = logger
self.writer = try AVAssetWriter(outputURL: outputURL, fileType: .mp4)
let settings: [String: Any] = [
AVVideoCodecKey: AVVideoCodecType.h264,
AVVideoWidthKey: width,
AVVideoHeightKey: height,
]
self.input = AVAssetWriterInput(mediaType: .video, outputSettings: settings)
self.input.expectsMediaDataInRealTime = true
guard self.writer.canAdd(self.input) else {
throw ScreenRecordService.ScreenRecordError.writeFailed("Cannot add video input")
}
self.writer.add(self.input)
if includeAudio {
let audioSettings: [String: Any] = [
AVFormatIDKey: kAudioFormatMPEG4AAC,
AVNumberOfChannelsKey: 1,
AVSampleRateKey: 44100,
AVEncoderBitRateKey: 96000,
]
let audioInput = AVAssetWriterInput(mediaType: .audio, outputSettings: audioSettings)
audioInput.expectsMediaDataInRealTime = true
if self.writer.canAdd(audioInput) {
self.writer.add(audioInput)
self.audioInput = audioInput
self.hasAudio = true
} else {
self.audioInput = nil
self.hasAudio = false
}
} else {
self.audioInput = nil
self.hasAudio = false
}
super.init()
}
func stream(_ stream: SCStream, didStopWithError error: any Error) {
self.queue.async {
let msg = String(describing: error)
self.pendingErrorMessage = msg
self.logger.error("screen record stream stopped with error: \(msg, privacy: .public)")
_ = stream
}
}
func stream(
_ stream: SCStream,
didOutputSampleBuffer sampleBuffer: CMSampleBuffer,
of type: SCStreamOutputType)
{
guard CMSampleBufferDataIsReady(sampleBuffer) else { return }
// Callback runs on `sampleHandlerQueue` (`self.queue`).
switch type {
case .screen:
self.handleVideo(sampleBuffer: sampleBuffer)
case .audio:
self.handleAudio(sampleBuffer: sampleBuffer)
case .microphone:
break
@unknown default:
break
}
_ = stream
}
private func handleVideo(sampleBuffer: CMSampleBuffer) {
if let msg = self.pendingErrorMessage {
self.logger.error("screen record aborting due to prior error: \(msg, privacy: .public)")
return
}
if self.didFinish { return }
if !self.started {
guard self.writer.startWriting() else {
self.pendingErrorMessage = self.writer.error?.localizedDescription ?? "Failed to start writer"
return
}
let pts = CMSampleBufferGetPresentationTimeStamp(sampleBuffer)
self.writer.startSession(atSourceTime: pts)
self.started = true
}
self.sawFrame = true
if self.input.isReadyForMoreMediaData {
_ = self.input.append(sampleBuffer)
}
}
private func handleAudio(sampleBuffer: CMSampleBuffer) {
guard let audioInput else { return }
if let msg = self.pendingErrorMessage {
self.logger.error("screen record audio aborting due to prior error: \(msg, privacy: .public)")
return
}
if self.didFinish || !self.started { return }
if audioInput.isReadyForMoreMediaData {
_ = audioInput.append(sampleBuffer)
}
}
func finish() async throws {
try await withCheckedThrowingContinuation { (cont: CheckedContinuation<Void, Error>) in
self.queue.async {
if let msg = self.pendingErrorMessage {
cont.resume(throwing: ScreenRecordService.ScreenRecordError.writeFailed(msg))
return
}
guard self.started, self.sawFrame else {
cont.resume(throwing: ScreenRecordService.ScreenRecordError.noFramesCaptured)
return
}
if self.didFinish {
cont.resume()
return
}
self.didFinish = true
self.input.markAsFinished()
self.audioInput?.markAsFinished()
self.writer.finishWriting {
if let err = self.writer.error {
cont
.resume(throwing: ScreenRecordService.ScreenRecordError
.writeFailed(err.localizedDescription))
} else if self.writer.status != .completed {
cont
.resume(throwing: ScreenRecordService.ScreenRecordError
.writeFailed("Failed to finalize video"))
} else {
cont.resume()
}
}
}
}
}
}

View File

@@ -0,0 +1,495 @@
import MoltbotChatUI
import MoltbotKit
import MoltbotProtocol
import OSLog
import SwiftUI
struct SessionPreviewItem: Identifiable, Sendable {
let id: String
let role: PreviewRole
let text: String
}
enum PreviewRole: String, Sendable {
case user
case assistant
case tool
case system
case other
var label: String {
switch self {
case .user: "User"
case .assistant: "Agent"
case .tool: "Tool"
case .system: "System"
case .other: "Other"
}
}
}
actor SessionPreviewCache {
static let shared = SessionPreviewCache()
private struct CacheEntry {
let snapshot: SessionMenuPreviewSnapshot
let updatedAt: Date
}
private var entries: [String: CacheEntry] = [:]
func cachedSnapshot(for sessionKey: String, maxAge: TimeInterval) -> SessionMenuPreviewSnapshot? {
guard let entry = self.entries[sessionKey] else { return nil }
guard Date().timeIntervalSince(entry.updatedAt) < maxAge else { return nil }
return entry.snapshot
}
func store(snapshot: SessionMenuPreviewSnapshot, for sessionKey: String) {
self.entries[sessionKey] = CacheEntry(snapshot: snapshot, updatedAt: Date())
}
func lastSnapshot(for sessionKey: String) -> SessionMenuPreviewSnapshot? {
self.entries[sessionKey]?.snapshot
}
}
actor SessionPreviewLimiter {
static let shared = SessionPreviewLimiter(maxConcurrent: 2)
private let maxConcurrent: Int
private var available: Int
private var waitQueue: [UUID] = []
private var waiters: [UUID: CheckedContinuation<Void, Never>] = [:]
init(maxConcurrent: Int) {
let normalized = max(1, maxConcurrent)
self.maxConcurrent = normalized
self.available = normalized
}
func withPermit<T>(_ operation: () async throws -> T) async throws -> T {
await self.acquire()
defer { self.release() }
if Task.isCancelled { throw CancellationError() }
return try await operation()
}
private func acquire() async {
if self.available > 0 {
self.available -= 1
return
}
let id = UUID()
await withCheckedContinuation { cont in
self.waitQueue.append(id)
self.waiters[id] = cont
}
}
private func release() {
if let id = self.waitQueue.first {
self.waitQueue.removeFirst()
if let cont = self.waiters.removeValue(forKey: id) {
cont.resume()
}
return
}
self.available = min(self.available + 1, self.maxConcurrent)
}
}
#if DEBUG
extension SessionPreviewCache {
func _testSet(
snapshot: SessionMenuPreviewSnapshot,
for sessionKey: String,
updatedAt: Date = Date())
{
self.entries[sessionKey] = CacheEntry(snapshot: snapshot, updatedAt: updatedAt)
}
func _testReset() {
self.entries = [:]
}
}
#endif
struct SessionMenuPreviewSnapshot: Sendable {
let items: [SessionPreviewItem]
let status: SessionMenuPreviewView.LoadStatus
}
struct SessionMenuPreviewView: View {
let width: CGFloat
let maxLines: Int
let title: String
let items: [SessionPreviewItem]
let status: LoadStatus
@Environment(\.menuItemHighlighted) private var isHighlighted
enum LoadStatus: Equatable {
case loading
case ready
case empty
case error(String)
}
private var primaryColor: Color {
if self.isHighlighted {
return Color(nsColor: .selectedMenuItemTextColor)
}
return Color(nsColor: .labelColor)
}
private var secondaryColor: Color {
if self.isHighlighted {
return Color(nsColor: .selectedMenuItemTextColor).opacity(0.85)
}
return Color(nsColor: .secondaryLabelColor)
}
var body: some View {
VStack(alignment: .leading, spacing: 8) {
HStack(alignment: .firstTextBaseline, spacing: 4) {
Text(self.title)
.font(.caption.weight(.semibold))
.foregroundStyle(self.secondaryColor)
Spacer(minLength: 8)
}
switch self.status {
case .loading:
self.placeholder("Loading preview…")
case .empty:
self.placeholder("No recent messages")
case let .error(message):
self.placeholder(message)
case .ready:
if self.items.isEmpty {
self.placeholder("No recent messages")
} else {
VStack(alignment: .leading, spacing: 6) {
ForEach(self.items) { item in
self.previewRow(item)
}
}
}
}
}
.padding(.vertical, 6)
.padding(.leading, 16)
.padding(.trailing, 11)
.frame(width: max(1, self.width), alignment: .leading)
}
@ViewBuilder
private func previewRow(_ item: SessionPreviewItem) -> some View {
HStack(alignment: .top, spacing: 4) {
Text(item.role.label)
.font(.caption2.monospacedDigit())
.foregroundStyle(self.roleColor(item.role))
.frame(width: 50, alignment: .leading)
Text(item.text)
.font(.caption)
.foregroundStyle(self.primaryColor)
.multilineTextAlignment(.leading)
.lineLimit(self.maxLines)
.truncationMode(.tail)
.fixedSize(horizontal: false, vertical: true)
}
}
private func roleColor(_ role: PreviewRole) -> Color {
if self.isHighlighted { return Color(nsColor: .selectedMenuItemTextColor).opacity(0.9) }
switch role {
case .user: return .accentColor
case .assistant: return .secondary
case .tool: return .orange
case .system: return .gray
case .other: return .secondary
}
}
@ViewBuilder
private func placeholder(_ text: String) -> some View {
Text(text)
.font(.caption)
.foregroundStyle(self.primaryColor)
}
}
enum SessionMenuPreviewLoader {
private static let logger = Logger(subsystem: "bot.molt", category: "SessionPreview")
private static let previewTimeoutSeconds: Double = 4
private static let cacheMaxAgeSeconds: TimeInterval = 30
private static let previewMaxChars = 240
private struct PreviewTimeoutError: LocalizedError {
var errorDescription: String? { "preview timeout" }
}
static func prewarm(sessionKeys: [String], maxItems: Int) async {
let keys = self.uniqueKeys(sessionKeys)
guard !keys.isEmpty else { return }
do {
let payload = try await self.requestPreview(keys: keys, maxItems: maxItems)
await self.cache(payload: payload, maxItems: maxItems)
} catch {
if self.isUnknownMethodError(error) { return }
let errorDescription = String(describing: error)
Self.logger.debug(
"Session preview prewarm failed count=\(keys.count, privacy: .public) " +
"error=\(errorDescription, privacy: .public)")
}
}
static func load(sessionKey: String, maxItems: Int) async -> SessionMenuPreviewSnapshot {
if let cached = await SessionPreviewCache.shared.cachedSnapshot(
for: sessionKey,
maxAge: cacheMaxAgeSeconds)
{
return cached
}
do {
let snapshot = try await self.fetchSnapshot(sessionKey: sessionKey, maxItems: maxItems)
await SessionPreviewCache.shared.store(snapshot: snapshot, for: sessionKey)
return snapshot
} catch is CancellationError {
return SessionMenuPreviewSnapshot(items: [], status: .loading)
} catch {
if let fallback = await SessionPreviewCache.shared.lastSnapshot(for: sessionKey) {
return fallback
}
let errorDescription = String(describing: error)
Self.logger.warning(
"Session preview failed session=\(sessionKey, privacy: .public) " +
"error=\(errorDescription, privacy: .public)")
return SessionMenuPreviewSnapshot(items: [], status: .error("Preview unavailable"))
}
}
private static func fetchSnapshot(sessionKey: String, maxItems: Int) async throws -> SessionMenuPreviewSnapshot {
do {
let payload = try await self.requestPreview(keys: [sessionKey], maxItems: maxItems)
if let entry = payload.previews.first(where: { $0.key == sessionKey }) ?? payload.previews.first {
return self.snapshot(from: entry, maxItems: maxItems)
}
return SessionMenuPreviewSnapshot(items: [], status: .error("Preview unavailable"))
} catch {
if self.isUnknownMethodError(error) {
return try await self.fetchHistorySnapshot(sessionKey: sessionKey, maxItems: maxItems)
}
throw error
}
}
private static func requestPreview(
keys: [String],
maxItems: Int) async throws -> MoltbotSessionsPreviewPayload
{
let boundedItems = self.normalizeMaxItems(maxItems)
let timeoutMs = Int(self.previewTimeoutSeconds * 1000)
return try await SessionPreviewLimiter.shared.withPermit {
try await AsyncTimeout.withTimeout(
seconds: self.previewTimeoutSeconds,
onTimeout: { PreviewTimeoutError() },
operation: {
try await GatewayConnection.shared.sessionsPreview(
keys: keys,
limit: boundedItems,
maxChars: self.previewMaxChars,
timeoutMs: timeoutMs)
})
}
}
private static func fetchHistorySnapshot(
sessionKey: String,
maxItems: Int) async throws -> SessionMenuPreviewSnapshot
{
let timeoutMs = Int(self.previewTimeoutSeconds * 1000)
let payload = try await SessionPreviewLimiter.shared.withPermit {
try await AsyncTimeout.withTimeout(
seconds: self.previewTimeoutSeconds,
onTimeout: { PreviewTimeoutError() },
operation: {
try await GatewayConnection.shared.chatHistory(
sessionKey: sessionKey,
limit: self.previewLimit(for: maxItems),
timeoutMs: timeoutMs)
})
}
let built = Self.previewItems(from: payload, maxItems: maxItems)
return Self.snapshot(from: built)
}
private static func snapshot(from items: [SessionPreviewItem]) -> SessionMenuPreviewSnapshot {
SessionMenuPreviewSnapshot(items: items, status: items.isEmpty ? .empty : .ready)
}
private static func snapshot(
from entry: MoltbotSessionPreviewEntry,
maxItems: Int) -> SessionMenuPreviewSnapshot
{
let items = self.previewItems(from: entry, maxItems: maxItems)
let normalized = entry.status.trimmingCharacters(in: .whitespacesAndNewlines).lowercased()
switch normalized {
case "ok":
return SessionMenuPreviewSnapshot(items: items, status: items.isEmpty ? .empty : .ready)
case "empty":
return SessionMenuPreviewSnapshot(items: items, status: .empty)
case "missing":
return SessionMenuPreviewSnapshot(items: items, status: .error("Session missing"))
default:
return SessionMenuPreviewSnapshot(items: items, status: .error("Preview unavailable"))
}
}
private static func cache(payload: MoltbotSessionsPreviewPayload, maxItems: Int) async {
for entry in payload.previews {
let snapshot = self.snapshot(from: entry, maxItems: maxItems)
await SessionPreviewCache.shared.store(snapshot: snapshot, for: entry.key)
}
}
private static func previewLimit(for maxItems: Int) -> Int {
let boundedItems = self.normalizeMaxItems(maxItems)
return min(max(boundedItems * 3, 20), 120)
}
private static func normalizeMaxItems(_ maxItems: Int) -> Int {
max(1, min(maxItems, 50))
}
private static func previewItems(
from entry: MoltbotSessionPreviewEntry,
maxItems: Int) -> [SessionPreviewItem]
{
let boundedItems = self.normalizeMaxItems(maxItems)
let built: [SessionPreviewItem] = entry.items.enumerated().compactMap { index, item in
let text = item.text.trimmingCharacters(in: .whitespacesAndNewlines)
guard !text.isEmpty else { return nil }
let role = self.previewRoleFromRaw(item.role)
return SessionPreviewItem(id: "\(entry.key)-\(index)", role: role, text: text)
}
let trimmed = built.suffix(boundedItems)
return Array(trimmed.reversed())
}
private static func previewItems(
from payload: MoltbotChatHistoryPayload,
maxItems: Int) -> [SessionPreviewItem]
{
let boundedItems = self.normalizeMaxItems(maxItems)
let raw: [MoltbotKit.AnyCodable] = payload.messages ?? []
let messages = self.decodeMessages(raw)
let built = messages.compactMap { message -> SessionPreviewItem? in
guard let text = self.previewText(for: message) else { return nil }
let isTool = self.isToolCall(message)
let role = self.previewRole(message.role, isTool: isTool)
let id = "\(message.timestamp ?? 0)-\(UUID().uuidString)"
return SessionPreviewItem(id: id, role: role, text: text)
}
let trimmed = built.suffix(boundedItems)
return Array(trimmed.reversed())
}
private static func decodeMessages(_ raw: [MoltbotKit.AnyCodable]) -> [MoltbotChatMessage] {
raw.compactMap { item in
guard let data = try? JSONEncoder().encode(item) else { return nil }
return try? JSONDecoder().decode(MoltbotChatMessage.self, from: data)
}
}
private static func previewRole(_ raw: String, isTool: Bool) -> PreviewRole {
if isTool { return .tool }
return self.previewRoleFromRaw(raw)
}
private static func previewRoleFromRaw(_ raw: String) -> PreviewRole {
switch raw.lowercased() {
case "user": .user
case "assistant": .assistant
case "system": .system
case "tool": .tool
default: .other
}
}
private static func previewText(for message: MoltbotChatMessage) -> String? {
let text = message.content.compactMap(\.text).joined(separator: "\n")
.trimmingCharacters(in: .whitespacesAndNewlines)
if !text.isEmpty { return text }
let toolNames = self.toolNames(for: message)
if !toolNames.isEmpty {
let shown = toolNames.prefix(2)
let overflow = toolNames.count - shown.count
var label = "call \(shown.joined(separator: ", "))"
if overflow > 0 { label += " +\(overflow)" }
return label
}
if let media = self.mediaSummary(for: message) {
return media
}
return nil
}
private static func isToolCall(_ message: MoltbotChatMessage) -> Bool {
if message.toolName?.nonEmpty != nil { return true }
return message.content.contains { $0.name?.nonEmpty != nil || $0.type?.lowercased() == "toolcall" }
}
private static func toolNames(for message: MoltbotChatMessage) -> [String] {
var names: [String] = []
for content in message.content {
if let name = content.name?.nonEmpty {
names.append(name)
}
}
if let toolName = message.toolName?.nonEmpty {
names.append(toolName)
}
return Self.dedupePreservingOrder(names)
}
private static func mediaSummary(for message: MoltbotChatMessage) -> String? {
let types = message.content.compactMap { content -> String? in
let raw = content.type?.trimmingCharacters(in: .whitespacesAndNewlines).lowercased()
guard let raw, !raw.isEmpty else { return nil }
if raw == "text" || raw == "toolcall" { return nil }
return raw
}
guard let first = types.first else { return nil }
return "[\(first)]"
}
private static func dedupePreservingOrder(_ values: [String]) -> [String] {
var seen = Set<String>()
var result: [String] = []
for value in values where !seen.contains(value) {
seen.insert(value)
result.append(value)
}
return result
}
private static func uniqueKeys(_ keys: [String]) -> [String] {
let trimmed = keys.map { $0.trimmingCharacters(in: .whitespacesAndNewlines) }
return self.dedupePreservingOrder(trimmed.filter { !$0.isEmpty })
}
private static func isUnknownMethodError(_ error: Error) -> Bool {
guard let response = error as? GatewayResponseError else { return false }
guard response.code == ErrorCode.invalidRequest.rawValue else { return false }
let message = response.message.lowercased()
return message.contains("unknown method")
}
}

View File

@@ -0,0 +1,226 @@
import AppKit
import Foundation
import Observation
import os
#if canImport(Darwin)
import Darwin
#endif
/// Manages Tailscale integration and status checking.
@Observable
@MainActor
final class TailscaleService {
static let shared = TailscaleService()
/// Tailscale local API endpoint.
private static let tailscaleAPIEndpoint = "http://100.100.100.100/api/data"
/// API request timeout in seconds.
private static let apiTimeoutInterval: TimeInterval = 5.0
private let logger = Logger(subsystem: "bot.molt", category: "tailscale")
/// Indicates if the Tailscale app is installed on the system.
private(set) var isInstalled = false
/// Indicates if Tailscale is currently running.
private(set) var isRunning = false
/// The Tailscale hostname for this device (e.g., "my-mac.tailnet.ts.net").
private(set) var tailscaleHostname: String?
/// The Tailscale IPv4 address for this device.
private(set) var tailscaleIP: String?
/// Error message if status check fails.
private(set) var statusError: String?
private init() {
Task { await self.checkTailscaleStatus() }
}
#if DEBUG
init(
isInstalled: Bool,
isRunning: Bool,
tailscaleHostname: String? = nil,
tailscaleIP: String? = nil,
statusError: String? = nil)
{
self.isInstalled = isInstalled
self.isRunning = isRunning
self.tailscaleHostname = tailscaleHostname
self.tailscaleIP = tailscaleIP
self.statusError = statusError
}
#endif
func checkAppInstallation() -> Bool {
let installed = FileManager().fileExists(atPath: "/Applications/Tailscale.app")
self.logger.info("Tailscale app installed: \(installed)")
return installed
}
private struct TailscaleAPIResponse: Codable {
let status: String
let deviceName: String
let tailnetName: String
let iPv4: String?
private enum CodingKeys: String, CodingKey {
case status = "Status"
case deviceName = "DeviceName"
case tailnetName = "TailnetName"
case iPv4 = "IPv4"
}
}
private func fetchTailscaleStatus() async -> TailscaleAPIResponse? {
guard let url = URL(string: Self.tailscaleAPIEndpoint) else {
self.logger.error("Invalid Tailscale API URL")
return nil
}
do {
let configuration = URLSessionConfiguration.default
configuration.timeoutIntervalForRequest = Self.apiTimeoutInterval
let session = URLSession(configuration: configuration)
let (data, response) = try await session.data(from: url)
guard let httpResponse = response as? HTTPURLResponse,
httpResponse.statusCode == 200
else {
self.logger.warning("Tailscale API returned non-200 status")
return nil
}
let decoder = JSONDecoder()
return try decoder.decode(TailscaleAPIResponse.self, from: data)
} catch {
self.logger.debug("Failed to fetch Tailscale status: \(String(describing: error))")
return nil
}
}
func checkTailscaleStatus() async {
let previousIP = self.tailscaleIP
self.isInstalled = self.checkAppInstallation()
if !self.isInstalled {
self.isRunning = false
self.tailscaleHostname = nil
self.tailscaleIP = nil
self.statusError = "Tailscale is not installed"
} else if let apiResponse = await fetchTailscaleStatus() {
self.isRunning = apiResponse.status.lowercased() == "running"
if self.isRunning {
let deviceName = apiResponse.deviceName
.lowercased()
.replacingOccurrences(of: " ", with: "-")
let tailnetName = apiResponse.tailnetName
.replacingOccurrences(of: ".ts.net", with: "")
.replacingOccurrences(of: ".tailscale.net", with: "")
self.tailscaleHostname = "\(deviceName).\(tailnetName).ts.net"
self.tailscaleIP = apiResponse.iPv4
self.statusError = nil
self.logger.info(
"Tailscale running host=\(self.tailscaleHostname ?? "nil") ip=\(self.tailscaleIP ?? "nil")")
} else {
self.tailscaleHostname = nil
self.tailscaleIP = nil
self.statusError = "Tailscale is not running"
}
} else {
self.isRunning = false
self.tailscaleHostname = nil
self.tailscaleIP = nil
self.statusError = "Please start the Tailscale app"
self.logger.info("Tailscale API not responding; app likely not running")
}
if self.tailscaleIP == nil, let fallback = Self.detectTailnetIPv4() {
self.tailscaleIP = fallback
if !self.isRunning {
self.isRunning = true
}
self.statusError = nil
self.logger.info("Tailscale interface IP detected (fallback) ip=\(fallback, privacy: .public)")
}
if previousIP != self.tailscaleIP {
await GatewayEndpointStore.shared.refresh()
}
}
func openTailscaleApp() {
if let url = URL(string: "file:///Applications/Tailscale.app") {
NSWorkspace.shared.open(url)
}
}
func openAppStore() {
if let url = URL(string: "https://apps.apple.com/us/app/tailscale/id1475387142") {
NSWorkspace.shared.open(url)
}
}
func openDownloadPage() {
if let url = URL(string: "https://tailscale.com/download/macos") {
NSWorkspace.shared.open(url)
}
}
func openSetupGuide() {
if let url = URL(string: "https://tailscale.com/kb/1017/install/") {
NSWorkspace.shared.open(url)
}
}
private nonisolated static func isTailnetIPv4(_ address: String) -> Bool {
let parts = address.split(separator: ".")
guard parts.count == 4 else { return false }
let octets = parts.compactMap { Int($0) }
guard octets.count == 4 else { return false }
let a = octets[0]
let b = octets[1]
return a == 100 && b >= 64 && b <= 127
}
private nonisolated static func detectTailnetIPv4() -> String? {
var addrList: UnsafeMutablePointer<ifaddrs>?
guard getifaddrs(&addrList) == 0, let first = addrList else { return nil }
defer { freeifaddrs(addrList) }
for ptr in sequence(first: first, next: { $0.pointee.ifa_next }) {
let flags = Int32(ptr.pointee.ifa_flags)
let isUp = (flags & IFF_UP) != 0
let isLoopback = (flags & IFF_LOOPBACK) != 0
let family = ptr.pointee.ifa_addr.pointee.sa_family
if !isUp || isLoopback || family != UInt8(AF_INET) { continue }
var addr = ptr.pointee.ifa_addr.pointee
var buffer = [CChar](repeating: 0, count: Int(NI_MAXHOST))
let result = getnameinfo(
&addr,
socklen_t(ptr.pointee.ifa_addr.pointee.sa_len),
&buffer,
socklen_t(buffer.count),
nil,
0,
NI_NUMERICHOST)
guard result == 0 else { continue }
let len = buffer.prefix { $0 != 0 }
let bytes = len.map { UInt8(bitPattern: $0) }
guard let ip = String(bytes: bytes, encoding: .utf8) else { continue }
if Self.isTailnetIPv4(ip) { return ip }
}
return nil
}
nonisolated static func fallbackTailnetIPv4() -> String? {
self.detectTailnetIPv4()
}
}

View File

@@ -0,0 +1,158 @@
import AVFoundation
import Foundation
import OSLog
@MainActor
final class TalkAudioPlayer: NSObject, @preconcurrency AVAudioPlayerDelegate {
static let shared = TalkAudioPlayer()
private let logger = Logger(subsystem: "bot.molt", category: "talk.tts")
private var player: AVAudioPlayer?
private var playback: Playback?
private final class Playback: @unchecked Sendable {
private let lock = NSLock()
private var finished = false
private var continuation: CheckedContinuation<TalkPlaybackResult, Never>?
private var watchdog: Task<Void, Never>?
func setContinuation(_ continuation: CheckedContinuation<TalkPlaybackResult, Never>) {
self.lock.lock()
defer { self.lock.unlock() }
self.continuation = continuation
}
func setWatchdog(_ task: Task<Void, Never>?) {
self.lock.lock()
let old = self.watchdog
self.watchdog = task
self.lock.unlock()
old?.cancel()
}
func cancelWatchdog() {
self.setWatchdog(nil)
}
func finish(_ result: TalkPlaybackResult) {
let continuation: CheckedContinuation<TalkPlaybackResult, Never>?
self.lock.lock()
if self.finished {
continuation = nil
} else {
self.finished = true
continuation = self.continuation
self.continuation = nil
}
self.lock.unlock()
continuation?.resume(returning: result)
}
}
func play(data: Data) async -> TalkPlaybackResult {
self.stopInternal()
let playback = Playback()
self.playback = playback
return await withCheckedContinuation { continuation in
playback.setContinuation(continuation)
do {
let player = try AVAudioPlayer(data: data)
self.player = player
player.delegate = self
player.prepareToPlay()
self.armWatchdog(playback: playback)
let ok = player.play()
if !ok {
self.logger.error("talk audio player refused to play")
self.finish(playback: playback, result: TalkPlaybackResult(finished: false, interruptedAt: nil))
}
} catch {
self.logger.error("talk audio player failed: \(error.localizedDescription, privacy: .public)")
self.finish(playback: playback, result: TalkPlaybackResult(finished: false, interruptedAt: nil))
}
}
}
func stop() -> Double? {
guard let player else { return nil }
let time = player.currentTime
self.stopInternal(interruptedAt: time)
return time
}
func audioPlayerDidFinishPlaying(_: AVAudioPlayer, successfully flag: Bool) {
self.stopInternal(finished: flag)
}
private func stopInternal(finished: Bool = false, interruptedAt: Double? = nil) {
guard let playback else { return }
let result = TalkPlaybackResult(finished: finished, interruptedAt: interruptedAt)
self.finish(playback: playback, result: result)
}
private func finish(playback: Playback, result: TalkPlaybackResult) {
playback.cancelWatchdog()
playback.finish(result)
guard self.playback === playback else { return }
self.playback = nil
self.player?.stop()
self.player = nil
}
private func stopInternal() {
if let playback = self.playback {
let interruptedAt = self.player?.currentTime
self.finish(
playback: playback,
result: TalkPlaybackResult(finished: false, interruptedAt: interruptedAt))
return
}
self.player?.stop()
self.player = nil
}
private func armWatchdog(playback: Playback) {
playback.setWatchdog(Task { @MainActor [weak self] in
guard let self else { return }
do {
try await Task.sleep(nanoseconds: 650_000_000)
} catch {
return
}
if Task.isCancelled { return }
guard self.playback === playback else { return }
if self.player?.isPlaying != true {
self.logger.error("talk audio player did not start playing")
self.finish(playback: playback, result: TalkPlaybackResult(finished: false, interruptedAt: nil))
return
}
let duration = self.player?.duration ?? 0
let timeoutSeconds = min(max(2.0, duration + 2.0), 5 * 60.0)
do {
try await Task.sleep(nanoseconds: UInt64(timeoutSeconds * 1_000_000_000))
} catch {
return
}
if Task.isCancelled { return }
guard self.playback === playback else { return }
guard self.player?.isPlaying == true else { return }
self.logger.error("talk audio player watchdog fired")
self.finish(playback: playback, result: TalkPlaybackResult(finished: false, interruptedAt: nil))
})
}
}
struct TalkPlaybackResult: Sendable {
let finished: Bool
let interruptedAt: Double?
}

View File

@@ -0,0 +1,69 @@
import Observation
@MainActor
@Observable
final class TalkModeController {
static let shared = TalkModeController()
private let logger = Logger(subsystem: "bot.molt", category: "talk.controller")
private(set) var phase: TalkModePhase = .idle
private(set) var isPaused: Bool = false
func setEnabled(_ enabled: Bool) async {
self.logger.info("talk enabled=\(enabled)")
if enabled {
TalkOverlayController.shared.present()
} else {
TalkOverlayController.shared.dismiss()
}
await TalkModeRuntime.shared.setEnabled(enabled)
}
func updatePhase(_ phase: TalkModePhase) {
self.phase = phase
TalkOverlayController.shared.updatePhase(phase)
let effectivePhase = self.isPaused ? "paused" : phase.rawValue
Task {
await GatewayConnection.shared.talkMode(
enabled: AppStateStore.shared.talkEnabled,
phase: effectivePhase)
}
}
func updateLevel(_ level: Double) {
TalkOverlayController.shared.updateLevel(level)
}
func setPaused(_ paused: Bool) {
guard self.isPaused != paused else { return }
self.logger.info("talk paused=\(paused)")
self.isPaused = paused
TalkOverlayController.shared.updatePaused(paused)
let effectivePhase = paused ? "paused" : self.phase.rawValue
Task {
await GatewayConnection.shared.talkMode(
enabled: AppStateStore.shared.talkEnabled,
phase: effectivePhase)
}
Task { await TalkModeRuntime.shared.setPaused(paused) }
}
func togglePaused() {
self.setPaused(!self.isPaused)
}
func stopSpeaking(reason: TalkStopReason = .userTap) {
Task { await TalkModeRuntime.shared.stopSpeaking(reason: reason) }
}
func exitTalkMode() {
Task { await AppStateStore.shared.setTalkEnabled(false) }
}
}
enum TalkStopReason {
case userTap
case speech
case manual
}

View File

@@ -0,0 +1,953 @@
import AVFoundation
import MoltbotChatUI
import MoltbotKit
import Foundation
import OSLog
import Speech
actor TalkModeRuntime {
static let shared = TalkModeRuntime()
private let logger = Logger(subsystem: "bot.molt", category: "talk.runtime")
private let ttsLogger = Logger(subsystem: "bot.molt", category: "talk.tts")
private static let defaultModelIdFallback = "eleven_v3"
private final class RMSMeter: @unchecked Sendable {
private let lock = NSLock()
private var latestRMS: Double = 0
func set(_ rms: Double) {
self.lock.lock()
self.latestRMS = rms
self.lock.unlock()
}
func get() -> Double {
self.lock.lock()
let value = self.latestRMS
self.lock.unlock()
return value
}
}
private var recognizer: SFSpeechRecognizer?
private var audioEngine: AVAudioEngine?
private var recognitionRequest: SFSpeechAudioBufferRecognitionRequest?
private var recognitionTask: SFSpeechRecognitionTask?
private var recognitionGeneration: Int = 0
private var rmsTask: Task<Void, Never>?
private let rmsMeter = RMSMeter()
private var captureTask: Task<Void, Never>?
private var silenceTask: Task<Void, Never>?
private var phase: TalkModePhase = .idle
private var isEnabled = false
private var isPaused = false
private var lifecycleGeneration: Int = 0
private var lastHeard: Date?
private var noiseFloorRMS: Double = 1e-4
private var lastTranscript: String = ""
private var lastSpeechEnergyAt: Date?
private var defaultVoiceId: String?
private var currentVoiceId: String?
private var defaultModelId: String?
private var currentModelId: String?
private var voiceOverrideActive = false
private var modelOverrideActive = false
private var defaultOutputFormat: String?
private var interruptOnSpeech: Bool = true
private var lastInterruptedAtSeconds: Double?
private var voiceAliases: [String: String] = [:]
private var lastSpokenText: String?
private var apiKey: String?
private var fallbackVoiceId: String?
private var lastPlaybackWasPCM: Bool = false
private let silenceWindow: TimeInterval = 0.7
private let minSpeechRMS: Double = 1e-3
private let speechBoostFactor: Double = 6.0
// MARK: - Lifecycle
func setEnabled(_ enabled: Bool) async {
guard enabled != self.isEnabled else { return }
self.isEnabled = enabled
self.lifecycleGeneration &+= 1
if enabled {
await self.start()
} else {
await self.stop()
}
}
func setPaused(_ paused: Bool) async {
guard paused != self.isPaused else { return }
self.isPaused = paused
await MainActor.run { TalkModeController.shared.updateLevel(0) }
guard self.isEnabled else { return }
if paused {
self.lastTranscript = ""
self.lastHeard = nil
self.lastSpeechEnergyAt = nil
await self.stopRecognition()
return
}
if self.phase == .idle || self.phase == .listening {
await self.startRecognition()
self.phase = .listening
await MainActor.run { TalkModeController.shared.updatePhase(.listening) }
self.startSilenceMonitor()
}
}
private func isCurrent(_ generation: Int) -> Bool {
generation == self.lifecycleGeneration && self.isEnabled
}
private func start() async {
let gen = self.lifecycleGeneration
guard voiceWakeSupported else { return }
guard PermissionManager.voiceWakePermissionsGranted() else {
self.logger.debug("talk runtime not starting: permissions missing")
return
}
await self.reloadConfig()
guard self.isCurrent(gen) else { return }
if self.isPaused {
self.phase = .idle
await MainActor.run {
TalkModeController.shared.updateLevel(0)
TalkModeController.shared.updatePhase(.idle)
}
return
}
await self.startRecognition()
guard self.isCurrent(gen) else { return }
self.phase = .listening
await MainActor.run { TalkModeController.shared.updatePhase(.listening) }
self.startSilenceMonitor()
}
private func stop() async {
self.captureTask?.cancel()
self.captureTask = nil
self.silenceTask?.cancel()
self.silenceTask = nil
// Stop audio before changing phase (stopSpeaking is gated on .speaking).
await self.stopSpeaking(reason: .manual)
self.lastTranscript = ""
self.lastHeard = nil
self.lastSpeechEnergyAt = nil
self.phase = .idle
await self.stopRecognition()
await MainActor.run {
TalkModeController.shared.updateLevel(0)
TalkModeController.shared.updatePhase(.idle)
}
}
// MARK: - Speech recognition
private struct RecognitionUpdate {
let transcript: String?
let hasConfidence: Bool
let isFinal: Bool
let errorDescription: String?
let generation: Int
}
private func startRecognition() async {
await self.stopRecognition()
self.recognitionGeneration &+= 1
let generation = self.recognitionGeneration
let locale = await MainActor.run { AppStateStore.shared.voiceWakeLocaleID }
self.recognizer = SFSpeechRecognizer(locale: Locale(identifier: locale))
guard let recognizer, recognizer.isAvailable else {
self.logger.error("talk recognizer unavailable")
return
}
self.recognitionRequest = SFSpeechAudioBufferRecognitionRequest()
self.recognitionRequest?.shouldReportPartialResults = true
guard let request = self.recognitionRequest else { return }
if self.audioEngine == nil {
self.audioEngine = AVAudioEngine()
}
guard let audioEngine = self.audioEngine else { return }
let input = audioEngine.inputNode
let format = input.outputFormat(forBus: 0)
input.removeTap(onBus: 0)
let meter = self.rmsMeter
input.installTap(onBus: 0, bufferSize: 2048, format: format) { [weak request, meter] buffer, _ in
request?.append(buffer)
if let rms = Self.rmsLevel(buffer: buffer) {
meter.set(rms)
}
}
audioEngine.prepare()
do {
try audioEngine.start()
} catch {
self.logger.error("talk audio engine start failed: \(error.localizedDescription, privacy: .public)")
return
}
self.startRMSTicker(meter: meter)
self.recognitionTask = recognizer.recognitionTask(with: request) { [weak self, generation] result, error in
guard let self else { return }
let segments = result?.bestTranscription.segments ?? []
let transcript = result?.bestTranscription.formattedString
let update = RecognitionUpdate(
transcript: transcript,
hasConfidence: segments.contains { $0.confidence > 0.6 },
isFinal: result?.isFinal ?? false,
errorDescription: error?.localizedDescription,
generation: generation)
Task { await self.handleRecognition(update) }
}
}
private func stopRecognition() async {
self.recognitionGeneration &+= 1
self.recognitionTask?.cancel()
self.recognitionTask = nil
self.recognitionRequest?.endAudio()
self.recognitionRequest = nil
self.audioEngine?.inputNode.removeTap(onBus: 0)
self.audioEngine?.stop()
self.audioEngine = nil
self.recognizer = nil
self.rmsTask?.cancel()
self.rmsTask = nil
}
private func startRMSTicker(meter: RMSMeter) {
self.rmsTask?.cancel()
self.rmsTask = Task { [weak self, meter] in
while let self {
try? await Task.sleep(nanoseconds: 50_000_000)
if Task.isCancelled { return }
await self.noteAudioLevel(rms: meter.get())
}
}
}
private func handleRecognition(_ update: RecognitionUpdate) async {
guard update.generation == self.recognitionGeneration else { return }
guard !self.isPaused else { return }
if let errorDescription = update.errorDescription {
self.logger.debug("talk recognition error: \(errorDescription, privacy: .public)")
}
guard let transcript = update.transcript else { return }
let trimmed = transcript.trimmingCharacters(in: .whitespacesAndNewlines)
if self.phase == .speaking, self.interruptOnSpeech {
if await self.shouldInterrupt(transcript: trimmed, hasConfidence: update.hasConfidence) {
await self.stopSpeaking(reason: .speech)
self.lastTranscript = ""
self.lastHeard = nil
await self.startListening()
}
return
}
guard self.phase == .listening else { return }
if !trimmed.isEmpty {
self.lastTranscript = trimmed
self.lastHeard = Date()
}
if update.isFinal {
self.lastTranscript = trimmed
}
}
// MARK: - Silence handling
private func startSilenceMonitor() {
self.silenceTask?.cancel()
self.silenceTask = Task { [weak self] in
await self?.silenceLoop()
}
}
private func silenceLoop() async {
while self.isEnabled {
try? await Task.sleep(nanoseconds: 200_000_000)
await self.checkSilence()
}
}
private func checkSilence() async {
guard !self.isPaused else { return }
guard self.phase == .listening else { return }
let transcript = self.lastTranscript.trimmingCharacters(in: .whitespacesAndNewlines)
guard !transcript.isEmpty else { return }
guard let lastHeard else { return }
let elapsed = Date().timeIntervalSince(lastHeard)
guard elapsed >= self.silenceWindow else { return }
await self.finalizeTranscript(transcript)
}
private func startListening() async {
self.phase = .listening
self.lastTranscript = ""
self.lastHeard = nil
await MainActor.run {
TalkModeController.shared.updatePhase(.listening)
TalkModeController.shared.updateLevel(0)
}
}
private func finalizeTranscript(_ text: String) async {
self.lastTranscript = ""
self.lastHeard = nil
self.phase = .thinking
await MainActor.run { TalkModeController.shared.updatePhase(.thinking) }
await self.stopRecognition()
await self.sendAndSpeak(text)
}
// MARK: - Gateway + TTS
private func sendAndSpeak(_ transcript: String) async {
let gen = self.lifecycleGeneration
await self.reloadConfig()
guard self.isCurrent(gen) else { return }
let prompt = self.buildPrompt(transcript: transcript)
let activeSessionKey = await MainActor.run { WebChatManager.shared.activeSessionKey }
let sessionKey: String = if let activeSessionKey {
activeSessionKey
} else {
await GatewayConnection.shared.mainSessionKey()
}
let runId = UUID().uuidString
let startedAt = Date().timeIntervalSince1970
self.logger.info(
"talk send start runId=\(runId, privacy: .public) " +
"session=\(sessionKey, privacy: .public) " +
"chars=\(prompt.count, privacy: .public)")
do {
let response = try await GatewayConnection.shared.chatSend(
sessionKey: sessionKey,
message: prompt,
thinking: "low",
idempotencyKey: runId,
attachments: [])
guard self.isCurrent(gen) else { return }
self.logger.info(
"talk chat.send ok runId=\(response.runId, privacy: .public) " +
"session=\(sessionKey, privacy: .public)")
guard let assistantText = await self.waitForAssistantText(
sessionKey: sessionKey,
since: startedAt,
timeoutSeconds: 45)
else {
self.logger.warning("talk assistant text missing after timeout")
await self.startListening()
await self.startRecognition()
return
}
guard self.isCurrent(gen) else { return }
self.logger.info("talk assistant text len=\(assistantText.count, privacy: .public)")
await self.playAssistant(text: assistantText)
guard self.isCurrent(gen) else { return }
await self.resumeListeningIfNeeded()
return
} catch {
self.logger.error("talk chat.send failed: \(error.localizedDescription, privacy: .public)")
await self.resumeListeningIfNeeded()
return
}
}
private func resumeListeningIfNeeded() async {
if self.isPaused {
self.lastTranscript = ""
self.lastHeard = nil
self.lastSpeechEnergyAt = nil
await MainActor.run {
TalkModeController.shared.updateLevel(0)
}
return
}
await self.startListening()
await self.startRecognition()
}
private func buildPrompt(transcript: String) -> String {
let interrupted = self.lastInterruptedAtSeconds
self.lastInterruptedAtSeconds = nil
return TalkPromptBuilder.build(transcript: transcript, interruptedAtSeconds: interrupted)
}
private func waitForAssistantText(
sessionKey: String,
since: Double,
timeoutSeconds: Int) async -> String?
{
let deadline = Date().addingTimeInterval(TimeInterval(timeoutSeconds))
while Date() < deadline {
if let text = await self.latestAssistantText(sessionKey: sessionKey, since: since) {
return text
}
try? await Task.sleep(nanoseconds: 300_000_000)
}
return nil
}
private func latestAssistantText(sessionKey: String, since: Double? = nil) async -> String? {
do {
let history = try await GatewayConnection.shared.chatHistory(sessionKey: sessionKey)
let messages = history.messages ?? []
let decoded: [MoltbotChatMessage] = messages.compactMap { item in
guard let data = try? JSONEncoder().encode(item) else { return nil }
return try? JSONDecoder().decode(MoltbotChatMessage.self, from: data)
}
let assistant = decoded.last { message in
guard message.role == "assistant" else { return false }
guard let since else { return true }
guard let timestamp = message.timestamp else { return false }
return TalkHistoryTimestamp.isAfter(timestamp, sinceSeconds: since)
}
guard let assistant else { return nil }
let text = assistant.content.compactMap(\.text).joined(separator: "\n")
let trimmed = text.trimmingCharacters(in: CharacterSet.whitespacesAndNewlines)
return trimmed.isEmpty ? nil : trimmed
} catch {
self.logger.error("talk history fetch failed: \(error.localizedDescription, privacy: .public)")
return nil
}
}
private func playAssistant(text: String) async {
guard let input = await self.preparePlaybackInput(text: text) else { return }
do {
if let apiKey = input.apiKey, !apiKey.isEmpty, let voiceId = input.voiceId {
try await self.playElevenLabs(input: input, apiKey: apiKey, voiceId: voiceId)
} else {
try await self.playSystemVoice(input: input)
}
} catch {
self.ttsLogger
.error(
"talk TTS failed: \(error.localizedDescription, privacy: .public); " +
"falling back to system voice")
do {
try await self.playSystemVoice(input: input)
} catch {
self.ttsLogger.error("talk system voice failed: \(error.localizedDescription, privacy: .public)")
}
}
if self.phase == .speaking {
self.phase = .thinking
await MainActor.run { TalkModeController.shared.updatePhase(.thinking) }
}
}
private struct TalkPlaybackInput {
let generation: Int
let cleanedText: String
let directive: TalkDirective?
let apiKey: String?
let voiceId: String?
let language: String?
let synthTimeoutSeconds: Double
}
private func preparePlaybackInput(text: String) async -> TalkPlaybackInput? {
let gen = self.lifecycleGeneration
let parse = TalkDirectiveParser.parse(text)
let directive = parse.directive
let cleaned = parse.stripped.trimmingCharacters(in: .whitespacesAndNewlines)
guard !cleaned.isEmpty else { return nil }
guard self.isCurrent(gen) else { return nil }
if !parse.unknownKeys.isEmpty {
self.logger
.warning(
"talk directive ignored keys: " +
"\(parse.unknownKeys.joined(separator: ","), privacy: .public)")
}
let requestedVoice = directive?.voiceId?.trimmingCharacters(in: .whitespacesAndNewlines)
let resolvedVoice = self.resolveVoiceAlias(requestedVoice)
if let requestedVoice, !requestedVoice.isEmpty, resolvedVoice == nil {
self.logger.warning("talk unknown voice alias \(requestedVoice, privacy: .public)")
}
if let voice = resolvedVoice {
if directive?.once == true {
self.logger.info("talk voice override (once) voiceId=\(voice, privacy: .public)")
} else {
self.currentVoiceId = voice
self.voiceOverrideActive = true
self.logger.info("talk voice override voiceId=\(voice, privacy: .public)")
}
}
if let model = directive?.modelId {
if directive?.once == true {
self.logger.info("talk model override (once) modelId=\(model, privacy: .public)")
} else {
self.currentModelId = model
self.modelOverrideActive = true
}
}
let apiKey = self.apiKey?.trimmingCharacters(in: .whitespacesAndNewlines)
let preferredVoice =
resolvedVoice ??
self.currentVoiceId ??
self.defaultVoiceId
let language = ElevenLabsTTSClient.validatedLanguage(directive?.language)
let voiceId: String? = if let apiKey, !apiKey.isEmpty {
await self.resolveVoiceId(preferred: preferredVoice, apiKey: apiKey)
} else {
nil
}
if apiKey?.isEmpty != false {
self.ttsLogger.warning("talk missing ELEVENLABS_API_KEY; falling back to system voice")
} else if voiceId == nil {
self.ttsLogger.warning("talk missing voiceId; falling back to system voice")
} else if let voiceId {
self.ttsLogger
.info(
"talk TTS request voiceId=\(voiceId, privacy: .public) " +
"chars=\(cleaned.count, privacy: .public)")
}
self.lastSpokenText = cleaned
let synthTimeoutSeconds = max(20.0, min(90.0, Double(cleaned.count) * 0.12))
guard self.isCurrent(gen) else { return nil }
return TalkPlaybackInput(
generation: gen,
cleanedText: cleaned,
directive: directive,
apiKey: apiKey,
voiceId: voiceId,
language: language,
synthTimeoutSeconds: synthTimeoutSeconds)
}
private func playElevenLabs(input: TalkPlaybackInput, apiKey: String, voiceId: String) async throws {
let desiredOutputFormat = input.directive?.outputFormat ?? self.defaultOutputFormat ?? "pcm_44100"
let outputFormat = ElevenLabsTTSClient.validatedOutputFormat(desiredOutputFormat)
if outputFormat == nil, !desiredOutputFormat.isEmpty {
self.logger
.warning(
"talk output_format unsupported for local playback: " +
"\(desiredOutputFormat, privacy: .public)")
}
let modelId = input.directive?.modelId ?? self.currentModelId ?? self.defaultModelId
func makeRequest(outputFormat: String?) -> ElevenLabsTTSRequest {
ElevenLabsTTSRequest(
text: input.cleanedText,
modelId: modelId,
outputFormat: outputFormat,
speed: TalkTTSValidation.resolveSpeed(
speed: input.directive?.speed,
rateWPM: input.directive?.rateWPM),
stability: TalkTTSValidation.validatedStability(
input.directive?.stability,
modelId: modelId),
similarity: TalkTTSValidation.validatedUnit(input.directive?.similarity),
style: TalkTTSValidation.validatedUnit(input.directive?.style),
speakerBoost: input.directive?.speakerBoost,
seed: TalkTTSValidation.validatedSeed(input.directive?.seed),
normalize: ElevenLabsTTSClient.validatedNormalize(input.directive?.normalize),
language: input.language,
latencyTier: TalkTTSValidation.validatedLatencyTier(input.directive?.latencyTier))
}
let request = makeRequest(outputFormat: outputFormat)
self.ttsLogger.info("talk TTS synth timeout=\(input.synthTimeoutSeconds, privacy: .public)s")
let client = ElevenLabsTTSClient(apiKey: apiKey)
let stream = client.streamSynthesize(voiceId: voiceId, request: request)
guard self.isCurrent(input.generation) else { return }
if self.interruptOnSpeech {
guard await self.prepareForPlayback(generation: input.generation) else { return }
}
await MainActor.run { TalkModeController.shared.updatePhase(.speaking) }
self.phase = .speaking
let result = await self.playRemoteStream(
client: client,
voiceId: voiceId,
outputFormat: outputFormat,
makeRequest: makeRequest,
stream: stream)
self.ttsLogger
.info(
"talk audio result finished=\(result.finished, privacy: .public) " +
"interruptedAt=\(String(describing: result.interruptedAt), privacy: .public)")
if !result.finished, result.interruptedAt == nil {
throw NSError(domain: "StreamingAudioPlayer", code: 1, userInfo: [
NSLocalizedDescriptionKey: "audio playback failed",
])
}
if !result.finished, let interruptedAt = result.interruptedAt, self.phase == .speaking {
if self.interruptOnSpeech {
self.lastInterruptedAtSeconds = interruptedAt
}
}
}
private func playRemoteStream(
client: ElevenLabsTTSClient,
voiceId: String,
outputFormat: String?,
makeRequest: (String?) -> ElevenLabsTTSRequest,
stream: AsyncThrowingStream<Data, Error>) async -> StreamingPlaybackResult
{
let sampleRate = TalkTTSValidation.pcmSampleRate(from: outputFormat)
if let sampleRate {
self.lastPlaybackWasPCM = true
let result = await self.playPCM(stream: stream, sampleRate: sampleRate)
if result.finished || result.interruptedAt != nil {
return result
}
let mp3Format = ElevenLabsTTSClient.validatedOutputFormat("mp3_44100")
self.ttsLogger.warning("talk pcm playback failed; retrying mp3")
self.lastPlaybackWasPCM = false
let mp3Stream = client.streamSynthesize(
voiceId: voiceId,
request: makeRequest(mp3Format))
return await self.playMP3(stream: mp3Stream)
}
self.lastPlaybackWasPCM = false
return await self.playMP3(stream: stream)
}
private func playSystemVoice(input: TalkPlaybackInput) async throws {
self.ttsLogger.info("talk system voice start chars=\(input.cleanedText.count, privacy: .public)")
if self.interruptOnSpeech {
guard await self.prepareForPlayback(generation: input.generation) else { return }
}
await MainActor.run { TalkModeController.shared.updatePhase(.speaking) }
self.phase = .speaking
await TalkSystemSpeechSynthesizer.shared.stop()
try await TalkSystemSpeechSynthesizer.shared.speak(
text: input.cleanedText,
language: input.language)
self.ttsLogger.info("talk system voice done")
}
private func prepareForPlayback(generation: Int) async -> Bool {
await self.startRecognition()
return self.isCurrent(generation)
}
private func resolveVoiceId(preferred: String?, apiKey: String) async -> String? {
let trimmed = preferred?.trimmingCharacters(in: .whitespacesAndNewlines) ?? ""
if !trimmed.isEmpty {
if let resolved = self.resolveVoiceAlias(trimmed) { return resolved }
self.ttsLogger.warning("talk unknown voice alias \(trimmed, privacy: .public)")
}
if let fallbackVoiceId { return fallbackVoiceId }
do {
let voices = try await ElevenLabsTTSClient(apiKey: apiKey).listVoices()
guard let first = voices.first else {
self.ttsLogger.error("elevenlabs voices list empty")
return nil
}
self.fallbackVoiceId = first.voiceId
if self.defaultVoiceId == nil {
self.defaultVoiceId = first.voiceId
}
if !self.voiceOverrideActive {
self.currentVoiceId = first.voiceId
}
let name = first.name ?? "unknown"
self.ttsLogger
.info("talk default voice selected \(name, privacy: .public) (\(first.voiceId, privacy: .public))")
return first.voiceId
} catch {
self.ttsLogger.error("elevenlabs list voices failed: \(error.localizedDescription, privacy: .public)")
return nil
}
}
private func resolveVoiceAlias(_ value: String?) -> String? {
let trimmed = (value ?? "").trimmingCharacters(in: .whitespacesAndNewlines)
guard !trimmed.isEmpty else { return nil }
let normalized = trimmed.lowercased()
if let mapped = self.voiceAliases[normalized] { return mapped }
if self.voiceAliases.values.contains(where: { $0.caseInsensitiveCompare(trimmed) == .orderedSame }) {
return trimmed
}
return Self.isLikelyVoiceId(trimmed) ? trimmed : nil
}
private static func isLikelyVoiceId(_ value: String) -> Bool {
guard value.count >= 10 else { return false }
return value.allSatisfy { $0.isLetter || $0.isNumber || $0 == "-" || $0 == "_" }
}
func stopSpeaking(reason: TalkStopReason) async {
let usePCM = self.lastPlaybackWasPCM
let interruptedAt = usePCM ? await self.stopPCM() : await self.stopMP3()
_ = usePCM ? await self.stopMP3() : await self.stopPCM()
await TalkSystemSpeechSynthesizer.shared.stop()
guard self.phase == .speaking else { return }
if reason == .speech, let interruptedAt {
self.lastInterruptedAtSeconds = interruptedAt
}
if reason == .manual {
return
}
if reason == .speech || reason == .userTap {
await self.startListening()
return
}
self.phase = .thinking
await MainActor.run { TalkModeController.shared.updatePhase(.thinking) }
}
}
extension TalkModeRuntime {
// MARK: - Audio playback (MainActor helpers)
@MainActor
private func playPCM(
stream: AsyncThrowingStream<Data, Error>,
sampleRate: Double) async -> StreamingPlaybackResult
{
await PCMStreamingAudioPlayer.shared.play(stream: stream, sampleRate: sampleRate)
}
@MainActor
private func playMP3(stream: AsyncThrowingStream<Data, Error>) async -> StreamingPlaybackResult {
await StreamingAudioPlayer.shared.play(stream: stream)
}
@MainActor
private func stopPCM() -> Double? {
PCMStreamingAudioPlayer.shared.stop()
}
@MainActor
private func stopMP3() -> Double? {
StreamingAudioPlayer.shared.stop()
}
// MARK: - Config
private func reloadConfig() async {
let cfg = await self.fetchTalkConfig()
self.defaultVoiceId = cfg.voiceId
self.voiceAliases = cfg.voiceAliases
if !self.voiceOverrideActive {
self.currentVoiceId = cfg.voiceId
}
self.defaultModelId = cfg.modelId
if !self.modelOverrideActive {
self.currentModelId = cfg.modelId
}
self.defaultOutputFormat = cfg.outputFormat
self.interruptOnSpeech = cfg.interruptOnSpeech
self.apiKey = cfg.apiKey
let hasApiKey = (cfg.apiKey?.isEmpty == false)
let voiceLabel = (cfg.voiceId?.isEmpty == false) ? cfg.voiceId! : "none"
let modelLabel = (cfg.modelId?.isEmpty == false) ? cfg.modelId! : "none"
self.logger
.info(
"talk config voiceId=\(voiceLabel, privacy: .public) " +
"modelId=\(modelLabel, privacy: .public) " +
"apiKey=\(hasApiKey, privacy: .public) " +
"interrupt=\(cfg.interruptOnSpeech, privacy: .public)")
}
private struct TalkRuntimeConfig {
let voiceId: String?
let voiceAliases: [String: String]
let modelId: String?
let outputFormat: String?
let interruptOnSpeech: Bool
let apiKey: String?
}
private func fetchTalkConfig() async -> TalkRuntimeConfig {
let env = ProcessInfo.processInfo.environment
let envVoice = env["ELEVENLABS_VOICE_ID"]?.trimmingCharacters(in: .whitespacesAndNewlines)
let sagVoice = env["SAG_VOICE_ID"]?.trimmingCharacters(in: .whitespacesAndNewlines)
let envApiKey = env["ELEVENLABS_API_KEY"]?.trimmingCharacters(in: .whitespacesAndNewlines)
do {
let snap: ConfigSnapshot = try await GatewayConnection.shared.requestDecoded(
method: .configGet,
params: nil,
timeoutMs: 8000)
let talk = snap.config?["talk"]?.dictionaryValue
let ui = snap.config?["ui"]?.dictionaryValue
let rawSeam = ui?["seamColor"]?.stringValue?.trimmingCharacters(in: .whitespacesAndNewlines) ?? ""
await MainActor.run {
AppStateStore.shared.seamColorHex = rawSeam.isEmpty ? nil : rawSeam
}
let voice = talk?["voiceId"]?.stringValue
let rawAliases = talk?["voiceAliases"]?.dictionaryValue
let resolvedAliases: [String: String] =
rawAliases?.reduce(into: [:]) { acc, entry in
let key = entry.key.trimmingCharacters(in: .whitespacesAndNewlines).lowercased()
let value = entry.value.stringValue?.trimmingCharacters(in: .whitespacesAndNewlines) ?? ""
guard !key.isEmpty, !value.isEmpty else { return }
acc[key] = value
} ?? [:]
let model = talk?["modelId"]?.stringValue?.trimmingCharacters(in: .whitespacesAndNewlines)
let resolvedModel = (model?.isEmpty == false) ? model! : Self.defaultModelIdFallback
let outputFormat = talk?["outputFormat"]?.stringValue
let interrupt = talk?["interruptOnSpeech"]?.boolValue
let apiKey = talk?["apiKey"]?.stringValue
let resolvedVoice =
(voice?.trimmingCharacters(in: .whitespacesAndNewlines).isEmpty == false ? voice : nil) ??
(envVoice?.isEmpty == false ? envVoice : nil) ??
(sagVoice?.isEmpty == false ? sagVoice : nil)
let resolvedApiKey =
(envApiKey?.isEmpty == false ? envApiKey : nil) ??
(apiKey?.trimmingCharacters(in: .whitespacesAndNewlines).isEmpty == false ? apiKey : nil)
return TalkRuntimeConfig(
voiceId: resolvedVoice,
voiceAliases: resolvedAliases,
modelId: resolvedModel,
outputFormat: outputFormat,
interruptOnSpeech: interrupt ?? true,
apiKey: resolvedApiKey)
} catch {
let resolvedVoice =
(envVoice?.isEmpty == false ? envVoice : nil) ??
(sagVoice?.isEmpty == false ? sagVoice : nil)
let resolvedApiKey = envApiKey?.isEmpty == false ? envApiKey : nil
return TalkRuntimeConfig(
voiceId: resolvedVoice,
voiceAliases: [:],
modelId: Self.defaultModelIdFallback,
outputFormat: nil,
interruptOnSpeech: true,
apiKey: resolvedApiKey)
}
}
// MARK: - Audio level handling
private func noteAudioLevel(rms: Double) async {
if self.phase != .listening, self.phase != .speaking { return }
let alpha: Double = rms < self.noiseFloorRMS ? 0.08 : 0.01
self.noiseFloorRMS = max(1e-7, self.noiseFloorRMS + (rms - self.noiseFloorRMS) * alpha)
let threshold = max(self.minSpeechRMS, self.noiseFloorRMS * self.speechBoostFactor)
if rms >= threshold {
let now = Date()
self.lastHeard = now
self.lastSpeechEnergyAt = now
}
if self.phase == .listening {
let clamped = min(1.0, max(0.0, rms / max(self.minSpeechRMS, threshold)))
await MainActor.run { TalkModeController.shared.updateLevel(clamped) }
}
}
private static func rmsLevel(buffer: AVAudioPCMBuffer) -> Double? {
guard let channelData = buffer.floatChannelData?.pointee else { return nil }
let frameCount = Int(buffer.frameLength)
guard frameCount > 0 else { return nil }
var sum: Double = 0
for i in 0..<frameCount {
let sample = Double(channelData[i])
sum += sample * sample
}
return sqrt(sum / Double(frameCount))
}
private func shouldInterrupt(transcript: String, hasConfidence: Bool) async -> Bool {
let trimmed = transcript.trimmingCharacters(in: .whitespacesAndNewlines)
guard trimmed.count >= 3 else { return false }
if self.isLikelyEcho(of: trimmed) { return false }
let now = Date()
if let lastSpeechEnergyAt, now.timeIntervalSince(lastSpeechEnergyAt) > 0.35 {
return false
}
return hasConfidence
}
private func isLikelyEcho(of transcript: String) -> Bool {
guard let spoken = self.lastSpokenText?.lowercased(), !spoken.isEmpty else { return false }
let probe = transcript.lowercased()
if probe.count < 6 {
return spoken.contains(probe)
}
return spoken.contains(probe)
}
private static func resolveSpeed(speed: Double?, rateWPM: Int?, logger: Logger) -> Double? {
if let rateWPM, rateWPM > 0 {
let resolved = Double(rateWPM) / 175.0
if resolved <= 0.5 || resolved >= 2.0 {
logger.warning("talk rateWPM out of range: \(rateWPM, privacy: .public)")
return nil
}
return resolved
}
if let speed {
if speed <= 0.5 || speed >= 2.0 {
logger.warning("talk speed out of range: \(speed, privacy: .public)")
return nil
}
return speed
}
return nil
}
private static func validatedUnit(_ value: Double?, name: String, logger: Logger) -> Double? {
guard let value else { return nil }
if value < 0 || value > 1 {
logger.warning("talk \(name, privacy: .public) out of range: \(value, privacy: .public)")
return nil
}
return value
}
private static func validatedSeed(_ value: Int?, logger: Logger) -> UInt32? {
guard let value else { return nil }
if value < 0 || value > 4_294_967_295 {
logger.warning("talk seed out of range: \(value, privacy: .public)")
return nil
}
return UInt32(value)
}
private static func validatedNormalize(_ value: String?, logger: Logger) -> String? {
guard let value else { return nil }
let normalized = value.trimmingCharacters(in: .whitespacesAndNewlines).lowercased()
guard ["auto", "on", "off"].contains(normalized) else {
logger.warning("talk normalize invalid: \(normalized, privacy: .public)")
return nil
}
return normalized
}
}

View File

@@ -0,0 +1,146 @@
import AppKit
import Observation
import OSLog
import SwiftUI
@MainActor
@Observable
final class TalkOverlayController {
static let shared = TalkOverlayController()
static let overlaySize: CGFloat = 440
static let orbSize: CGFloat = 96
static let orbPadding: CGFloat = 12
static let orbHitSlop: CGFloat = 10
private let logger = Logger(subsystem: "bot.molt", category: "talk.overlay")
struct Model {
var isVisible: Bool = false
var phase: TalkModePhase = .idle
var isPaused: Bool = false
var level: Double = 0
}
var model = Model()
private var window: NSPanel?
private var hostingView: NSHostingView<TalkOverlayView>?
private let screenInset: CGFloat = 0
func present() {
self.ensureWindow()
self.hostingView?.rootView = TalkOverlayView(controller: self)
let target = self.targetFrame()
guard let window else { return }
if !self.model.isVisible {
self.model.isVisible = true
let start = target.offsetBy(dx: 0, dy: -6)
window.setFrame(start, display: true)
window.alphaValue = 0
window.orderFrontRegardless()
NSAnimationContext.runAnimationGroup { context in
context.duration = 0.18
context.timingFunction = CAMediaTimingFunction(name: .easeOut)
window.animator().setFrame(target, display: true)
window.animator().alphaValue = 1
}
} else {
window.setFrame(target, display: true)
window.orderFrontRegardless()
}
}
func dismiss() {
guard let window else {
self.model.isVisible = false
return
}
let target = window.frame.offsetBy(dx: 6, dy: 6)
NSAnimationContext.runAnimationGroup { context in
context.duration = 0.16
context.timingFunction = CAMediaTimingFunction(name: .easeOut)
window.animator().setFrame(target, display: true)
window.animator().alphaValue = 0
} completionHandler: {
Task { @MainActor in
window.orderOut(nil)
self.model.isVisible = false
}
}
}
func updatePhase(_ phase: TalkModePhase) {
guard self.model.phase != phase else { return }
self.logger.info("talk overlay phase=\(phase.rawValue, privacy: .public)")
self.model.phase = phase
}
func updatePaused(_ paused: Bool) {
guard self.model.isPaused != paused else { return }
self.logger.info("talk overlay paused=\(paused)")
self.model.isPaused = paused
}
func updateLevel(_ level: Double) {
guard self.model.isVisible else { return }
self.model.level = max(0, min(1, level))
}
func currentWindowOrigin() -> CGPoint? {
self.window?.frame.origin
}
func setWindowOrigin(_ origin: CGPoint) {
guard let window else { return }
window.setFrameOrigin(origin)
}
// MARK: - Private
private func ensureWindow() {
if self.window != nil { return }
let panel = NSPanel(
contentRect: NSRect(x: 0, y: 0, width: Self.overlaySize, height: Self.overlaySize),
styleMask: [.nonactivatingPanel, .borderless],
backing: .buffered,
defer: false)
panel.isOpaque = false
panel.backgroundColor = .clear
panel.hasShadow = false
panel.level = NSWindow.Level(rawValue: NSWindow.Level.popUpMenu.rawValue - 4)
panel.collectionBehavior = [.canJoinAllSpaces, .fullScreenAuxiliary, .transient]
panel.hidesOnDeactivate = false
panel.isMovable = false
panel.acceptsMouseMovedEvents = true
panel.isFloatingPanel = true
panel.becomesKeyOnlyIfNeeded = true
panel.titleVisibility = .hidden
panel.titlebarAppearsTransparent = true
let host = TalkOverlayHostingView(rootView: TalkOverlayView(controller: self))
host.translatesAutoresizingMaskIntoConstraints = false
panel.contentView = host
self.hostingView = host
self.window = panel
}
private func targetFrame() -> NSRect {
let screen = self.window?.screen
?? NSScreen.main
?? NSScreen.screens.first
guard let screen else { return .zero }
let size = NSSize(width: Self.overlaySize, height: Self.overlaySize)
let visible = screen.visibleFrame
let origin = CGPoint(
x: visible.maxX - size.width - self.screenInset,
y: visible.maxY - size.height - self.screenInset)
return NSRect(origin: origin, size: size)
}
}
private final class TalkOverlayHostingView: NSHostingView<TalkOverlayView> {
override func acceptsFirstMouse(for event: NSEvent?) -> Bool {
true
}
}

View File

@@ -0,0 +1,53 @@
import AppKit
import Foundation
import OSLog
@MainActor
final class TerminationSignalWatcher {
static let shared = TerminationSignalWatcher()
private let logger = Logger(subsystem: "bot.molt", category: "lifecycle")
private var sources: [DispatchSourceSignal] = []
private var terminationRequested = false
func start() {
guard self.sources.isEmpty else { return }
self.install(SIGTERM)
self.install(SIGINT)
}
func stop() {
for s in self.sources {
s.cancel()
}
self.sources.removeAll(keepingCapacity: false)
self.terminationRequested = false
}
private func install(_ sig: Int32) {
// Make sure the default action doesn't kill the process before we can gracefully shut down.
signal(sig, SIG_IGN)
let source = DispatchSource.makeSignalSource(signal: sig, queue: .main)
source.setEventHandler { [weak self] in
self?.handle(sig)
}
source.resume()
self.sources.append(source)
}
private func handle(_ sig: Int32) {
guard !self.terminationRequested else { return }
self.terminationRequested = true
self.logger.info("received signal \(sig, privacy: .public); terminating")
// Ensure any pairing prompt can't accidentally approve during shutdown.
NodePairingApprovalPrompter.shared.stop()
DevicePairingApprovalPrompter.shared.stop()
NSApp.terminate(nil)
// Safety net: don't hang forever if something blocks termination.
DispatchQueue.main.asyncAfter(deadline: .now() + 3) {
exit(0)
}
}
}

View File

@@ -0,0 +1,421 @@
import AppKit
import AVFoundation
import Dispatch
import OSLog
import Speech
/// Observes right Option and starts a push-to-talk capture while it is held.
final class VoicePushToTalkHotkey: @unchecked Sendable {
static let shared = VoicePushToTalkHotkey()
private var globalMonitor: Any?
private var localMonitor: Any?
private var optionDown = false // right option only
private var active = false
private let beginAction: @Sendable () async -> Void
private let endAction: @Sendable () async -> Void
init(
beginAction: @escaping @Sendable () async -> Void = { await VoicePushToTalk.shared.begin() },
endAction: @escaping @Sendable () async -> Void = { await VoicePushToTalk.shared.end() })
{
self.beginAction = beginAction
self.endAction = endAction
}
func setEnabled(_ enabled: Bool) {
if ProcessInfo.processInfo.isRunningTests { return }
self.withMainThread { [weak self] in
guard let self else { return }
if enabled {
self.startMonitoring()
} else {
self.stopMonitoring()
}
}
}
private func startMonitoring() {
// assert(Thread.isMainThread) - Removed for Swift 6
guard self.globalMonitor == nil, self.localMonitor == nil else { return }
// Listen-only global monitor; we rely on Input Monitoring permission to receive events.
self.globalMonitor = NSEvent.addGlobalMonitorForEvents(matching: .flagsChanged) { [weak self] event in
let keyCode = event.keyCode
let flags = event.modifierFlags
self?.handleFlagsChanged(keyCode: keyCode, modifierFlags: flags)
}
// Also listen locally so we still catch events when the app is active/focused.
self.localMonitor = NSEvent.addLocalMonitorForEvents(matching: .flagsChanged) { [weak self] event in
let keyCode = event.keyCode
let flags = event.modifierFlags
self?.handleFlagsChanged(keyCode: keyCode, modifierFlags: flags)
return event
}
}
private func stopMonitoring() {
// assert(Thread.isMainThread) - Removed for Swift 6
if let globalMonitor {
NSEvent.removeMonitor(globalMonitor)
self.globalMonitor = nil
}
if let localMonitor {
NSEvent.removeMonitor(localMonitor)
self.localMonitor = nil
}
self.optionDown = false
self.active = false
}
private func handleFlagsChanged(keyCode: UInt16, modifierFlags: NSEvent.ModifierFlags) {
self.withMainThread { [weak self] in
self?.updateModifierState(keyCode: keyCode, modifierFlags: modifierFlags)
}
}
private func withMainThread(_ block: @escaping @Sendable () -> Void) {
DispatchQueue.main.async(execute: block)
}
private func updateModifierState(keyCode: UInt16, modifierFlags: NSEvent.ModifierFlags) {
// assert(Thread.isMainThread) - Removed for Swift 6
// Right Option (keyCode 61) acts as a hold-to-talk modifier.
if keyCode == 61 {
self.optionDown = modifierFlags.contains(.option)
}
let chordActive = self.optionDown
if chordActive, !self.active {
self.active = true
Task {
Logger(subsystem: "bot.molt", category: "voicewake.ptt")
.info("ptt hotkey down")
await self.beginAction()
}
} else if !chordActive, self.active {
self.active = false
Task {
Logger(subsystem: "bot.molt", category: "voicewake.ptt")
.info("ptt hotkey up")
await self.endAction()
}
}
}
func _testUpdateModifierState(keyCode: UInt16, modifierFlags: NSEvent.ModifierFlags) {
self.updateModifierState(keyCode: keyCode, modifierFlags: modifierFlags)
}
}
/// Short-lived speech recognizer that records while the hotkey is held.
actor VoicePushToTalk {
static let shared = VoicePushToTalk()
private let logger = Logger(subsystem: "bot.molt", category: "voicewake.ptt")
private var recognizer: SFSpeechRecognizer?
// Lazily created on begin() to avoid creating an AVAudioEngine at app launch, which can switch Bluetooth
// headphones into the low-quality headset profile even if push-to-talk is never used.
private var audioEngine: AVAudioEngine?
private var recognitionRequest: SFSpeechAudioBufferRecognitionRequest?
private var recognitionTask: SFSpeechRecognitionTask?
private var tapInstalled = false
// Session token used to drop stale callbacks when a new capture starts.
private var sessionID = UUID()
private var committed: String = ""
private var volatile: String = ""
private var activeConfig: Config?
private var isCapturing = false
private var triggerChimePlayed = false
private var finalized = false
private var timeoutTask: Task<Void, Never>?
private var overlayToken: UUID?
private var adoptedPrefix: String = ""
private struct Config {
let micID: String?
let localeID: String?
let triggerChime: VoiceWakeChime
let sendChime: VoiceWakeChime
}
func begin() async {
guard voiceWakeSupported else { return }
guard !self.isCapturing else { return }
// Start a fresh session and invalidate any in-flight callbacks tied to an older one.
let sessionID = UUID()
self.sessionID = sessionID
// Ensure permissions up front.
let granted = await PermissionManager.ensureVoiceWakePermissions(interactive: true)
guard granted else { return }
let config = await MainActor.run { self.makeConfig() }
self.activeConfig = config
self.isCapturing = true
self.triggerChimePlayed = false
self.finalized = false
self.timeoutTask?.cancel(); self.timeoutTask = nil
let snapshot = await MainActor.run { VoiceSessionCoordinator.shared.snapshot() }
self.adoptedPrefix = snapshot.visible ? snapshot.text.trimmingCharacters(in: .whitespacesAndNewlines) : ""
self.logger.info("ptt begin adopted_prefix_len=\(self.adoptedPrefix.count, privacy: .public)")
if config.triggerChime != .none {
self.triggerChimePlayed = true
await MainActor.run { VoiceWakeChimePlayer.play(config.triggerChime, reason: "ptt.trigger") }
}
// Pause the always-on wake word recognizer so both pipelines don't fight over the mic tap.
await VoiceWakeRuntime.shared.pauseForPushToTalk()
let adoptedPrefix = self.adoptedPrefix
let adoptedAttributed: NSAttributedString? = adoptedPrefix.isEmpty ? nil : Self.makeAttributed(
committed: adoptedPrefix,
volatile: "",
isFinal: false)
self.overlayToken = await MainActor.run {
VoiceSessionCoordinator.shared.startSession(
source: .pushToTalk,
text: adoptedPrefix,
attributed: adoptedAttributed,
forwardEnabled: true)
}
do {
try await self.startRecognition(localeID: config.localeID, sessionID: sessionID)
} catch {
await MainActor.run {
VoiceWakeOverlayController.shared.dismiss()
}
self.isCapturing = false
// If push-to-talk fails to start after pausing wake-word, ensure we resume listening.
await VoiceWakeRuntime.shared.applyPushToTalkCooldown()
await VoiceWakeRuntime.shared.refresh(state: AppStateStore.shared)
}
}
func end() async {
guard self.isCapturing else { return }
self.isCapturing = false
let sessionID = self.sessionID
// Stop feeding Speech buffers first, then end the request. Stopping the engine here can race with
// Speech draining its converter chain (and we already stop/cancel in finalize).
if self.tapInstalled {
self.audioEngine?.inputNode.removeTap(onBus: 0)
self.tapInstalled = false
}
self.recognitionRequest?.endAudio()
// If we captured nothing, dismiss immediately when the user lets go.
if self.committed.isEmpty, self.volatile.isEmpty, self.adoptedPrefix.isEmpty {
await self.finalize(transcriptOverride: "", reason: "emptyOnRelease", sessionID: sessionID)
return
}
// Otherwise, give Speech a brief window to deliver the final result; then fall back.
self.timeoutTask?.cancel()
self.timeoutTask = Task { [weak self] in
try? await Task.sleep(nanoseconds: 1_500_000_000) // 1.5s grace period to await final result
await self?.finalize(transcriptOverride: nil, reason: "timeout", sessionID: sessionID)
}
}
// MARK: - Private
private func startRecognition(localeID: String?, sessionID: UUID) async throws {
let locale = localeID.flatMap { Locale(identifier: $0) } ?? Locale(identifier: Locale.current.identifier)
self.recognizer = SFSpeechRecognizer(locale: locale)
guard let recognizer, recognizer.isAvailable else {
throw NSError(
domain: "VoicePushToTalk",
code: 1,
userInfo: [NSLocalizedDescriptionKey: "Recognizer unavailable"])
}
self.recognitionRequest = SFSpeechAudioBufferRecognitionRequest()
self.recognitionRequest?.shouldReportPartialResults = true
guard let request = self.recognitionRequest else { return }
// Lazily create the engine here so app launch doesn't grab audio resources / trigger Bluetooth HFP.
if self.audioEngine == nil {
self.audioEngine = AVAudioEngine()
}
guard let audioEngine = self.audioEngine else { return }
let input = audioEngine.inputNode
let format = input.outputFormat(forBus: 0)
if self.tapInstalled {
input.removeTap(onBus: 0)
self.tapInstalled = false
}
// Pipe raw mic buffers into the Speech request while the chord is held.
input.installTap(onBus: 0, bufferSize: 2048, format: format) { [weak request] buffer, _ in
request?.append(buffer)
}
self.tapInstalled = true
audioEngine.prepare()
try audioEngine.start()
self.recognitionTask = recognizer.recognitionTask(with: request) { [weak self] result, error in
guard let self else { return }
if let error {
self.logger.debug("push-to-talk error: \(error.localizedDescription, privacy: .public)")
}
let transcript = result?.bestTranscription.formattedString
let isFinal = result?.isFinal ?? false
// Hop to a Task so UI updates stay off the Speech callback thread.
Task.detached { [weak self, transcript, isFinal, sessionID] in
guard let self else { return }
await self.handle(transcript: transcript, isFinal: isFinal, sessionID: sessionID)
}
}
}
private func handle(transcript: String?, isFinal: Bool, sessionID: UUID) async {
guard sessionID == self.sessionID else {
self.logger.debug("push-to-talk drop transcript for stale session")
return
}
guard let transcript else { return }
if isFinal {
self.committed = transcript
self.volatile = ""
} else {
self.volatile = Self.delta(after: self.committed, current: transcript)
}
let committedWithPrefix = Self.join(self.adoptedPrefix, self.committed)
let snapshot = Self.join(committedWithPrefix, self.volatile)
let attributed = Self.makeAttributed(committed: committedWithPrefix, volatile: self.volatile, isFinal: isFinal)
if let token = self.overlayToken {
await MainActor.run {
VoiceSessionCoordinator.shared.updatePartial(
token: token,
text: snapshot,
attributed: attributed)
}
}
}
private func finalize(transcriptOverride: String?, reason: String, sessionID: UUID?) async {
if self.finalized { return }
if let sessionID, sessionID != self.sessionID {
self.logger.debug("push-to-talk drop finalize for stale session")
return
}
self.finalized = true
self.isCapturing = false
self.timeoutTask?.cancel(); self.timeoutTask = nil
let finalRecognized: String = {
if let override = transcriptOverride?.trimmingCharacters(in: .whitespacesAndNewlines) {
return override
}
return (self.committed + self.volatile).trimmingCharacters(in: .whitespacesAndNewlines)
}()
let finalText = Self.join(self.adoptedPrefix, finalRecognized)
let chime = finalText.isEmpty ? .none : (self.activeConfig?.sendChime ?? .none)
let token = self.overlayToken
let logger = self.logger
await MainActor.run {
logger.info("ptt finalize reason=\(reason, privacy: .public) len=\(finalText.count, privacy: .public)")
if let token {
VoiceSessionCoordinator.shared.finalize(
token: token,
text: finalText,
sendChime: chime,
autoSendAfter: nil)
VoiceSessionCoordinator.shared.sendNow(token: token, reason: reason)
} else if !finalText.isEmpty {
if chime != .none {
VoiceWakeChimePlayer.play(chime, reason: "ptt.fallback_send")
}
Task.detached {
await VoiceWakeForwarder.forward(transcript: finalText)
}
}
}
self.recognitionTask?.cancel()
self.recognitionRequest = nil
self.recognitionTask = nil
if self.tapInstalled {
self.audioEngine?.inputNode.removeTap(onBus: 0)
self.tapInstalled = false
}
if self.audioEngine?.isRunning == true {
self.audioEngine?.stop()
self.audioEngine?.reset()
}
// Release the engine so we also release any audio session/resources when push-to-talk ends.
self.audioEngine = nil
self.committed = ""
self.volatile = ""
self.activeConfig = nil
self.triggerChimePlayed = false
self.overlayToken = nil
self.adoptedPrefix = ""
// Resume the wake-word runtime after push-to-talk finishes.
await VoiceWakeRuntime.shared.applyPushToTalkCooldown()
_ = await MainActor.run { Task { await VoiceWakeRuntime.shared.refresh(state: AppStateStore.shared) } }
}
@MainActor
private func makeConfig() -> Config {
let state = AppStateStore.shared
return Config(
micID: state.voiceWakeMicID.isEmpty ? nil : state.voiceWakeMicID,
localeID: state.voiceWakeLocaleID,
triggerChime: state.voiceWakeTriggerChime,
sendChime: state.voiceWakeSendChime)
}
// MARK: - Test helpers
static func _testDelta(committed: String, current: String) -> String {
self.delta(after: committed, current: current)
}
static func _testAttributedColors(isFinal: Bool) -> (NSColor, NSColor) {
let sample = self.makeAttributed(committed: "a", volatile: "b", isFinal: isFinal)
let committedColor = sample.attribute(.foregroundColor, at: 0, effectiveRange: nil) as? NSColor ?? .clear
let volatileColor = sample.attribute(.foregroundColor, at: 1, effectiveRange: nil) as? NSColor ?? .clear
return (committedColor, volatileColor)
}
private static func join(_ prefix: String, _ suffix: String) -> String {
if prefix.isEmpty { return suffix }
if suffix.isEmpty { return prefix }
return "\(prefix) \(suffix)"
}
private static func delta(after committed: String, current: String) -> String {
if current.hasPrefix(committed) {
let start = current.index(current.startIndex, offsetBy: committed.count)
return String(current[start...])
}
return current
}
private static func makeAttributed(committed: String, volatile: String, isFinal: Bool) -> NSAttributedString {
let full = NSMutableAttributedString()
let committedAttr: [NSAttributedString.Key: Any] = [
.foregroundColor: NSColor.labelColor,
.font: NSFont.systemFont(ofSize: 13, weight: .regular),
]
full.append(NSAttributedString(string: committed, attributes: committedAttr))
let volatileColor: NSColor = isFinal ? .labelColor : NSColor.tertiaryLabelColor
let volatileAttr: [NSAttributedString.Key: Any] = [
.foregroundColor: volatileColor,
.font: NSFont.systemFont(ofSize: 13, weight: .regular),
]
full.append(NSAttributedString(string: volatile, attributes: volatileAttr))
return full
}
}

View File

@@ -0,0 +1,134 @@
import AppKit
import Foundation
import Observation
@MainActor
@Observable
final class VoiceSessionCoordinator {
static let shared = VoiceSessionCoordinator()
enum Source: String { case wakeWord, pushToTalk }
struct Session {
let token: UUID
let source: Source
var text: String
var attributed: NSAttributedString?
var isFinal: Bool
var sendChime: VoiceWakeChime
var autoSendDelay: TimeInterval?
}
private let logger = Logger(subsystem: "bot.molt", category: "voicewake.coordinator")
private var session: Session?
// MARK: - API
func startSession(
source: Source,
text: String,
attributed: NSAttributedString? = nil,
forwardEnabled: Bool = false) -> UUID
{
let token = UUID()
self.logger.info("coordinator start token=\(token.uuidString) source=\(source.rawValue) len=\(text.count)")
let attributedText = attributed ?? VoiceWakeOverlayController.shared.makeAttributed(from: text)
let session = Session(
token: token,
source: source,
text: text,
attributed: attributedText,
isFinal: false,
sendChime: .none,
autoSendDelay: nil)
self.session = session
VoiceWakeOverlayController.shared.startSession(
token: token,
source: VoiceWakeOverlayController.Source(rawValue: source.rawValue) ?? .wakeWord,
transcript: text,
attributed: attributedText,
forwardEnabled: forwardEnabled,
isFinal: false)
return token
}
func updatePartial(token: UUID, text: String, attributed: NSAttributedString? = nil) {
guard let session, session.token == token else { return }
self.session?.text = text
self.session?.attributed = attributed
VoiceWakeOverlayController.shared.updatePartial(token: token, transcript: text, attributed: attributed)
}
func finalize(
token: UUID,
text: String,
sendChime: VoiceWakeChime,
autoSendAfter: TimeInterval?)
{
guard let session, session.token == token else { return }
self.logger
.info(
"coordinator finalize token=\(token.uuidString) len=\(text.count) autoSendAfter=\(autoSendAfter ?? -1)")
self.session?.text = text
self.session?.isFinal = true
self.session?.sendChime = sendChime
self.session?.autoSendDelay = autoSendAfter
let attributed = VoiceWakeOverlayController.shared.makeAttributed(from: text)
VoiceWakeOverlayController.shared.presentFinal(
token: token,
transcript: text,
autoSendAfter: autoSendAfter,
sendChime: sendChime,
attributed: attributed)
}
func sendNow(token: UUID, reason: String = "explicit") {
guard let session, session.token == token else { return }
let text = session.text.trimmingCharacters(in: .whitespacesAndNewlines)
guard !text.isEmpty else {
self.logger.info("coordinator sendNow \(reason) empty -> dismiss")
VoiceWakeOverlayController.shared.dismiss(token: token, reason: .empty, outcome: .empty)
self.clearSession()
return
}
VoiceWakeOverlayController.shared.beginSendUI(token: token, sendChime: session.sendChime)
Task.detached {
_ = await VoiceWakeForwarder.forward(transcript: text)
}
}
func dismiss(
token: UUID,
reason: VoiceWakeOverlayController.DismissReason,
outcome: VoiceWakeOverlayController.SendOutcome)
{
guard let session, session.token == token else { return }
VoiceWakeOverlayController.shared.dismiss(token: token, reason: reason, outcome: outcome)
self.clearSession()
}
func updateLevel(token: UUID, _ level: Double) {
guard let session, session.token == token else { return }
VoiceWakeOverlayController.shared.updateLevel(token: token, level)
}
func snapshot() -> (token: UUID?, text: String, visible: Bool) {
(self.session?.token, self.session?.text ?? "", VoiceWakeOverlayController.shared.isVisible)
}
// MARK: - Private
private func clearSession() {
self.session = nil
}
/// Overlay dismiss completion callback (manual X, empty, auto-dismiss after send).
/// Ensures the wake-word recognizer is resumed if Voice Wake is enabled.
func overlayDidDismiss(token: UUID?) {
if let token, self.session?.token == token {
self.clearSession()
}
Task { await VoiceWakeRuntime.shared.refresh(state: AppStateStore.shared) }
}
}

View File

@@ -0,0 +1,74 @@
import AppKit
import Foundation
import OSLog
enum VoiceWakeChime: Codable, Equatable, Sendable {
case none
case system(name: String)
case custom(displayName: String, bookmark: Data)
var systemName: String? {
if case let .system(name) = self {
return name
}
return nil
}
var displayLabel: String {
switch self {
case .none:
"No Sound"
case let .system(name):
VoiceWakeChimeCatalog.displayName(for: name)
case let .custom(displayName, _):
displayName
}
}
}
enum VoiceWakeChimeCatalog {
/// Options shown in the picker.
static var systemOptions: [String] { SoundEffectCatalog.systemOptions }
static func displayName(for raw: String) -> String {
SoundEffectCatalog.displayName(for: raw)
}
static func url(for name: String) -> URL? {
SoundEffectCatalog.url(for: name)
}
}
@MainActor
enum VoiceWakeChimePlayer {
private static let logger = Logger(subsystem: "bot.molt", category: "voicewake.chime")
private static var lastSound: NSSound?
static func play(_ chime: VoiceWakeChime, reason: String? = nil) {
guard let sound = self.sound(for: chime) else { return }
if let reason {
self.logger.log(level: .info, "chime play reason=\(reason, privacy: .public)")
} else {
self.logger.log(level: .info, "chime play")
}
DiagnosticsFileLog.shared.log(category: "voicewake.chime", event: "play", fields: [
"reason": reason ?? "",
"chime": chime.displayLabel,
"systemName": chime.systemName ?? "",
])
SoundEffectPlayer.play(sound)
}
private static func sound(for chime: VoiceWakeChime) -> NSSound? {
switch chime {
case .none:
nil
case let .system(name):
SoundEffectPlayer.sound(named: name)
case let .custom(_, bookmark):
SoundEffectPlayer.sound(from: bookmark)
}
}
}

View File

@@ -0,0 +1,73 @@
import Foundation
import OSLog
enum VoiceWakeForwarder {
private static let logger = Logger(subsystem: "bot.molt", category: "voicewake.forward")
static func prefixedTranscript(_ transcript: String, machineName: String? = nil) -> String {
let resolvedMachine = machineName
.flatMap { name -> String? in
let trimmed = name.trimmingCharacters(in: .whitespacesAndNewlines)
return trimmed.isEmpty ? nil : trimmed
}
?? Host.current().localizedName
?? ProcessInfo.processInfo.hostName
let safeMachine = resolvedMachine.isEmpty ? "this Mac" : resolvedMachine
return """
User talked via voice recognition on \(safeMachine) - repeat prompt first \
+ remember some words might be incorrectly transcribed.
\(transcript)
"""
}
enum VoiceWakeForwardError: LocalizedError, Equatable {
case rpcFailed(String)
var errorDescription: String? {
switch self {
case let .rpcFailed(message): message
}
}
}
struct ForwardOptions: Sendable {
var sessionKey: String = "main"
var thinking: String = "low"
var deliver: Bool = true
var to: String?
var channel: GatewayAgentChannel = .last
}
@discardableResult
static func forward(
transcript: String,
options: ForwardOptions = ForwardOptions()) async -> Result<Void, VoiceWakeForwardError>
{
let payload = Self.prefixedTranscript(transcript)
let deliver = options.channel.shouldDeliver(options.deliver)
let result = await GatewayConnection.shared.sendAgent(GatewayAgentInvocation(
message: payload,
sessionKey: options.sessionKey,
thinking: options.thinking,
deliver: deliver,
to: options.to,
channel: options.channel))
if result.ok {
self.logger.info("voice wake forward ok")
return .success(())
}
let message = result.error ?? "agent rpc unavailable"
self.logger.error("voice wake forward failed: \(message, privacy: .public)")
return .failure(.rpcFailed(message))
}
static func checkConnection() async -> Result<Void, VoiceWakeForwardError> {
let status = await GatewayConnection.shared.status()
if status.ok { return .success(()) }
return .failure(.rpcFailed(status.error ?? "agent rpc unreachable"))
}
}

View File

@@ -0,0 +1,66 @@
import MoltbotKit
import Foundation
import OSLog
@MainActor
final class VoiceWakeGlobalSettingsSync {
static let shared = VoiceWakeGlobalSettingsSync()
private let logger = Logger(subsystem: "bot.molt", category: "voicewake.sync")
private var task: Task<Void, Never>?
private struct VoiceWakePayload: Codable, Equatable {
let triggers: [String]
}
func start() {
guard self.task == nil else { return }
self.task = Task { [weak self] in
guard let self else { return }
while !Task.isCancelled {
do {
try await GatewayConnection.shared.refresh()
} catch {
// Not configured / not reachable yet.
}
await self.refreshFromGateway()
let stream = await GatewayConnection.shared.subscribe(bufferingNewest: 200)
for await push in stream {
if Task.isCancelled { return }
await self.handle(push: push)
}
// If the stream finishes (gateway shutdown / reconnect), loop and resubscribe.
try? await Task.sleep(nanoseconds: 600_000_000)
}
}
}
func stop() {
self.task?.cancel()
self.task = nil
}
private func refreshFromGateway() async {
do {
let triggers = try await GatewayConnection.shared.voiceWakeGetTriggers()
AppStateStore.shared.applyGlobalVoiceWakeTriggers(triggers)
} catch {
// Best-effort only.
}
}
func handle(push: GatewayPush) async {
guard case let .event(evt) = push else { return }
guard evt.event == "voicewake.changed" else { return }
guard let payload = evt.payload else { return }
do {
let decoded = try GatewayPayloadDecoding.decode(payload, as: VoiceWakePayload.self)
AppStateStore.shared.applyGlobalVoiceWakeTriggers(decoded.triggers)
} catch {
self.logger.error("failed to decode voicewake.changed: \(error.localizedDescription, privacy: .public)")
}
}
}

View File

@@ -0,0 +1,60 @@
import AppKit
import Observation
import SwiftUI
/// Lightweight, borderless panel that shows the current voice wake transcript near the menu bar.
@MainActor
@Observable
final class VoiceWakeOverlayController {
static let shared = VoiceWakeOverlayController()
let logger = Logger(subsystem: "bot.molt", category: "voicewake.overlay")
let enableUI: Bool
/// Keep the voice wake overlay above any other Moltbot windows, but below the systems pop-up menus.
/// (Menu bar menus typically live at `.popUpMenu`.)
static let preferredWindowLevel = NSWindow.Level(rawValue: NSWindow.Level.popUpMenu.rawValue - 4)
enum Source: String { case wakeWord, pushToTalk }
var model = Model()
var isVisible: Bool { self.model.isVisible }
struct Model {
var text: String = ""
var isFinal: Bool = false
var isVisible: Bool = false
var forwardEnabled: Bool = false
var isSending: Bool = false
var attributed: NSAttributedString = .init(string: "")
var isOverflowing: Bool = false
var isEditing: Bool = false
var level: Double = 0 // normalized 0...1 speech level for UI
}
var window: NSPanel?
var hostingView: NSHostingView<VoiceWakeOverlayView>?
var autoSendTask: Task<Void, Never>?
var autoSendToken: UUID?
var activeToken: UUID?
var activeSource: Source?
var lastLevelUpdate: TimeInterval = 0
let width: CGFloat = 360
let padding: CGFloat = 10
let buttonWidth: CGFloat = 36
let spacing: CGFloat = 8
let verticalPadding: CGFloat = 8
let maxHeight: CGFloat = 400
let minHeight: CGFloat = 48
let closeOverflow: CGFloat = 10
let levelUpdateInterval: TimeInterval = 1.0 / 12.0
enum DismissReason { case explicit, empty }
enum SendOutcome { case sent, empty }
enum GuardOutcome { case accept, dropMismatch, dropNoActive }
init(enableUI: Bool = true) {
self.enableUI = enableUI
}
}

View File

@@ -0,0 +1,804 @@
import AVFoundation
import Foundation
import OSLog
import Speech
import SwabbleKit
#if canImport(AppKit)
import AppKit
#endif
/// Background listener that keeps the voice-wake pipeline alive outside the settings test view.
actor VoiceWakeRuntime {
static let shared = VoiceWakeRuntime()
enum ListeningState { case idle, voiceWake, pushToTalk }
private let logger = Logger(subsystem: "bot.molt", category: "voicewake.runtime")
private var recognizer: SFSpeechRecognizer?
// Lazily created on start to avoid creating an AVAudioEngine at app launch, which can switch Bluetooth
// headphones into the low-quality headset profile even if Voice Wake is disabled.
private var audioEngine: AVAudioEngine?
private var recognitionRequest: SFSpeechAudioBufferRecognitionRequest?
private var recognitionTask: SFSpeechRecognitionTask?
private var recognitionGeneration: Int = 0 // drop stale callbacks after restarts
private var lastHeard: Date?
private var noiseFloorRMS: Double = 1e-4
private var captureStartedAt: Date?
private var captureTask: Task<Void, Never>?
private var capturedTranscript: String = ""
private var isCapturing: Bool = false
private var heardBeyondTrigger: Bool = false
private var triggerChimePlayed: Bool = false
private var committedTranscript: String = ""
private var volatileTranscript: String = ""
private var cooldownUntil: Date?
private var currentConfig: RuntimeConfig?
private var listeningState: ListeningState = .idle
private var overlayToken: UUID?
private var activeTriggerEndTime: TimeInterval?
private var scheduledRestartTask: Task<Void, Never>?
private var lastLoggedText: String?
private var lastLoggedAt: Date?
private var lastTapLogAt: Date?
private var lastCallbackLogAt: Date?
private var lastTranscript: String?
private var lastTranscriptAt: Date?
private var preDetectTask: Task<Void, Never>?
private var isStarting: Bool = false
private var triggerOnlyTask: Task<Void, Never>?
// Tunables
// Silence threshold once we've captured user speech (post-trigger).
private let silenceWindow: TimeInterval = 2.0
// Silence threshold when we only heard the trigger but no post-trigger speech yet.
private let triggerOnlySilenceWindow: TimeInterval = 5.0
// Maximum capture duration from trigger until we force-send, to avoid runaway sessions.
private let captureHardStop: TimeInterval = 120.0
private let debounceAfterSend: TimeInterval = 0.35
// Voice activity detection parameters (RMS-based).
private let minSpeechRMS: Double = 1e-3
private let speechBoostFactor: Double = 6.0 // how far above noise floor we require to mark speech
private let preDetectSilenceWindow: TimeInterval = 1.0
private let triggerPauseWindow: TimeInterval = 0.55
/// Stops the active Speech pipeline without clearing the stored config, so we can restart cleanly.
private func haltRecognitionPipeline() {
// Bump generation first so any in-flight callbacks from the cancelled task get dropped.
self.recognitionGeneration &+= 1
self.recognitionTask?.cancel()
self.recognitionTask = nil
self.recognitionRequest?.endAudio()
self.recognitionRequest = nil
self.audioEngine?.inputNode.removeTap(onBus: 0)
self.audioEngine?.stop()
// Release the engine so we also release any audio session/resources when Voice Wake is idle.
self.audioEngine = nil
}
struct RuntimeConfig: Equatable {
let triggers: [String]
let micID: String?
let localeID: String?
let triggerChime: VoiceWakeChime
let sendChime: VoiceWakeChime
}
private struct RecognitionUpdate {
let transcript: String?
let segments: [WakeWordSegment]
let isFinal: Bool
let error: Error?
let generation: Int
}
func refresh(state: AppState) async {
let snapshot = await MainActor.run { () -> (Bool, RuntimeConfig) in
let enabled = state.swabbleEnabled
let config = RuntimeConfig(
triggers: sanitizeVoiceWakeTriggers(state.swabbleTriggerWords),
micID: state.voiceWakeMicID.isEmpty ? nil : state.voiceWakeMicID,
localeID: state.voiceWakeLocaleID.isEmpty ? nil : state.voiceWakeLocaleID,
triggerChime: state.voiceWakeTriggerChime,
sendChime: state.voiceWakeSendChime)
return (enabled, config)
}
guard voiceWakeSupported, snapshot.0 else {
self.stop()
return
}
guard PermissionManager.voiceWakePermissionsGranted() else {
self.logger.debug("voicewake runtime not starting: permissions missing")
self.stop()
return
}
let config = snapshot.1
if self.isStarting {
return
}
if self.scheduledRestartTask != nil, config == self.currentConfig, self.recognitionTask == nil {
return
}
if self.scheduledRestartTask != nil {
self.scheduledRestartTask?.cancel()
self.scheduledRestartTask = nil
}
if config == self.currentConfig, self.recognitionTask != nil {
return
}
self.stop()
await self.start(with: config)
}
private func start(with config: RuntimeConfig) async {
if self.isStarting {
return
}
self.isStarting = true
defer { self.isStarting = false }
do {
self.recognitionGeneration &+= 1
let generation = self.recognitionGeneration
self.configureSession(localeID: config.localeID)
guard let recognizer, recognizer.isAvailable else {
self.logger.error("voicewake runtime: speech recognizer unavailable")
return
}
self.recognitionRequest = SFSpeechAudioBufferRecognitionRequest()
self.recognitionRequest?.shouldReportPartialResults = true
self.recognitionRequest?.taskHint = .dictation
guard let request = self.recognitionRequest else { return }
// Lazily create the engine here so app launch doesn't grab audio resources / trigger Bluetooth HFP.
if self.audioEngine == nil {
self.audioEngine = AVAudioEngine()
}
guard let audioEngine = self.audioEngine else { return }
let input = audioEngine.inputNode
let format = input.outputFormat(forBus: 0)
guard format.channelCount > 0, format.sampleRate > 0 else {
throw NSError(
domain: "VoiceWakeRuntime",
code: 1,
userInfo: [NSLocalizedDescriptionKey: "No audio input available"])
}
input.removeTap(onBus: 0)
input.installTap(onBus: 0, bufferSize: 2048, format: format) { [weak self, weak request] buffer, _ in
request?.append(buffer)
guard let rms = Self.rmsLevel(buffer: buffer) else { return }
Task.detached { [weak self] in
await self?.noteAudioLevel(rms: rms)
await self?.noteAudioTap(rms: rms)
}
}
audioEngine.prepare()
try audioEngine.start()
self.currentConfig = config
self.lastHeard = Date()
// Preserve any existing cooldownUntil so the debounce after send isn't wiped by a restart.
self.recognitionTask = recognizer.recognitionTask(with: request) { [weak self, generation] result, error in
guard let self else { return }
let transcript = result?.bestTranscription.formattedString
let segments = result.flatMap { result in
transcript
.map { WakeWordSpeechSegments.from(transcription: result.bestTranscription, transcript: $0) }
} ?? []
let isFinal = result?.isFinal ?? false
Task { await self.noteRecognitionCallback(transcript: transcript, isFinal: isFinal, error: error) }
let update = RecognitionUpdate(
transcript: transcript,
segments: segments,
isFinal: isFinal,
error: error,
generation: generation)
Task { await self.handleRecognition(update, config: config) }
}
let preferred = config.micID?.isEmpty == false ? config.micID! : "system-default"
self.logger.info(
"voicewake runtime input preferred=\(preferred, privacy: .public) " +
"\(AudioInputDeviceObserver.defaultInputDeviceSummary(), privacy: .public)")
self.logger.info("voicewake runtime started")
DiagnosticsFileLog.shared.log(category: "voicewake.runtime", event: "started", fields: [
"locale": config.localeID ?? "",
"micID": config.micID ?? "",
])
} catch {
self.logger.error("voicewake runtime failed to start: \(error.localizedDescription, privacy: .public)")
self.stop()
}
}
private func stop(dismissOverlay: Bool = true, cancelScheduledRestart: Bool = true) {
if cancelScheduledRestart {
self.scheduledRestartTask?.cancel()
self.scheduledRestartTask = nil
}
self.captureTask?.cancel()
self.captureTask = nil
self.isCapturing = false
self.capturedTranscript = ""
self.captureStartedAt = nil
self.triggerChimePlayed = false
self.lastTranscript = nil
self.lastTranscriptAt = nil
self.preDetectTask?.cancel()
self.preDetectTask = nil
self.triggerOnlyTask?.cancel()
self.triggerOnlyTask = nil
self.haltRecognitionPipeline()
self.recognizer = nil
self.currentConfig = nil
self.listeningState = .idle
self.activeTriggerEndTime = nil
self.logger.debug("voicewake runtime stopped")
DiagnosticsFileLog.shared.log(category: "voicewake.runtime", event: "stopped")
let token = self.overlayToken
self.overlayToken = nil
guard dismissOverlay else { return }
Task { @MainActor in
if let token {
VoiceSessionCoordinator.shared.dismiss(token: token, reason: .explicit, outcome: .empty)
} else {
VoiceWakeOverlayController.shared.dismiss()
}
}
}
private func configureSession(localeID: String?) {
let locale = localeID.flatMap { Locale(identifier: $0) } ?? Locale(identifier: Locale.current.identifier)
self.recognizer = SFSpeechRecognizer(locale: locale)
self.recognizer?.defaultTaskHint = .dictation
}
private func handleRecognition(_ update: RecognitionUpdate, config: RuntimeConfig) async {
if update.generation != self.recognitionGeneration {
return // stale callback from a superseded recognizer session
}
if let error = update.error {
self.logger.debug("voicewake recognition error: \(error.localizedDescription, privacy: .public)")
}
guard let transcript = update.transcript else { return }
let now = Date()
if !transcript.isEmpty {
self.lastHeard = now
if !self.isCapturing {
self.lastTranscript = transcript
self.lastTranscriptAt = now
}
if self.isCapturing {
self.maybeLogRecognition(
transcript: transcript,
segments: update.segments,
triggers: config.triggers,
isFinal: update.isFinal,
match: nil,
usedFallback: false,
capturing: true)
let trimmed = Self.commandAfterTrigger(
transcript: transcript,
segments: update.segments,
triggerEndTime: self.activeTriggerEndTime,
triggers: config.triggers)
self.capturedTranscript = trimmed
self.updateHeardBeyondTrigger(withTrimmed: trimmed)
if update.isFinal {
self.committedTranscript = trimmed
self.volatileTranscript = ""
} else {
self.volatileTranscript = Self.delta(after: self.committedTranscript, current: trimmed)
}
let attributed = Self.makeAttributed(
committed: self.committedTranscript,
volatile: self.volatileTranscript,
isFinal: update.isFinal)
let snapshot = self.committedTranscript + self.volatileTranscript
if let token = self.overlayToken {
await MainActor.run {
VoiceSessionCoordinator.shared.updatePartial(
token: token,
text: snapshot,
attributed: attributed)
}
}
}
}
if self.isCapturing { return }
let gateConfig = WakeWordGateConfig(triggers: config.triggers)
var usedFallback = false
var match = WakeWordGate.match(transcript: transcript, segments: update.segments, config: gateConfig)
if match == nil, update.isFinal {
match = self.textOnlyFallbackMatch(
transcript: transcript,
triggers: config.triggers,
config: gateConfig)
usedFallback = match != nil
}
self.maybeLogRecognition(
transcript: transcript,
segments: update.segments,
triggers: config.triggers,
isFinal: update.isFinal,
match: match,
usedFallback: usedFallback,
capturing: false)
if let match {
if let cooldown = cooldownUntil, now < cooldown {
return
}
if usedFallback {
self.logger.info("voicewake runtime detected (text-only fallback) len=\(match.command.count)")
} else {
self.logger.info("voicewake runtime detected len=\(match.command.count)")
}
await self.beginCapture(command: match.command, triggerEndTime: match.triggerEndTime, config: config)
} else if !transcript.isEmpty, update.error == nil {
if self.isTriggerOnly(transcript: transcript, triggers: config.triggers) {
self.preDetectTask?.cancel()
self.preDetectTask = nil
self.scheduleTriggerOnlyPauseCheck(triggers: config.triggers, config: config)
} else {
self.triggerOnlyTask?.cancel()
self.triggerOnlyTask = nil
self.schedulePreDetectSilenceCheck(
triggers: config.triggers,
gateConfig: gateConfig,
config: config)
}
}
}
private func maybeLogRecognition(
transcript: String,
segments: [WakeWordSegment],
triggers: [String],
isFinal: Bool,
match: WakeWordGateMatch?,
usedFallback: Bool,
capturing: Bool)
{
guard !transcript.isEmpty else { return }
let level = self.logger.logLevel
guard level == .debug || level == .trace else { return }
if transcript == self.lastLoggedText, !isFinal {
if let last = self.lastLoggedAt, Date().timeIntervalSince(last) < 0.25 {
return
}
}
self.lastLoggedText = transcript
self.lastLoggedAt = Date()
let textOnly = WakeWordGate.matchesTextOnly(text: transcript, triggers: triggers)
let timingCount = segments.count(where: { $0.start > 0 || $0.duration > 0 })
let matchSummary = match.map {
"match=true gap=\(String(format: "%.2f", $0.postGap))s cmdLen=\($0.command.count)"
} ?? "match=false"
let segmentSummary = segments.map { seg in
let start = String(format: "%.2f", seg.start)
let end = String(format: "%.2f", seg.end)
return "\(seg.text)@\(start)-\(end)"
}.joined(separator: ", ")
self.logger.debug(
"voicewake runtime transcript='\(transcript, privacy: .private)' textOnly=\(textOnly) " +
"isFinal=\(isFinal) timing=\(timingCount)/\(segments.count) " +
"capturing=\(capturing) fallback=\(usedFallback) " +
"\(matchSummary) segments=[\(segmentSummary, privacy: .private)]")
}
private func noteAudioTap(rms: Double) {
let now = Date()
if let last = self.lastTapLogAt, now.timeIntervalSince(last) < 1.0 {
return
}
self.lastTapLogAt = now
let db = 20 * log10(max(rms, 1e-7))
self.logger.debug(
"voicewake runtime audio tap rms=\(String(format: "%.6f", rms)) " +
"db=\(String(format: "%.1f", db)) capturing=\(self.isCapturing)")
}
private func noteRecognitionCallback(transcript: String?, isFinal: Bool, error: Error?) {
guard transcript?.isEmpty ?? true else { return }
let now = Date()
if let last = self.lastCallbackLogAt, now.timeIntervalSince(last) < 1.0 {
return
}
self.lastCallbackLogAt = now
let errorSummary = error?.localizedDescription ?? "none"
self.logger.debug(
"voicewake runtime callback empty transcript isFinal=\(isFinal) error=\(errorSummary, privacy: .public)")
}
private func scheduleTriggerOnlyPauseCheck(triggers: [String], config: RuntimeConfig) {
self.triggerOnlyTask?.cancel()
let lastSeenAt = self.lastTranscriptAt
let lastText = self.lastTranscript
let windowNanos = UInt64(self.triggerPauseWindow * 1_000_000_000)
self.triggerOnlyTask = Task { [weak self, lastSeenAt, lastText] in
try? await Task.sleep(nanoseconds: windowNanos)
guard let self else { return }
await self.triggerOnlyPauseCheck(
lastSeenAt: lastSeenAt,
lastText: lastText,
triggers: triggers,
config: config)
}
}
private func schedulePreDetectSilenceCheck(
triggers: [String],
gateConfig: WakeWordGateConfig,
config: RuntimeConfig)
{
self.preDetectTask?.cancel()
let lastSeenAt = self.lastTranscriptAt
let lastText = self.lastTranscript
let windowNanos = UInt64(self.preDetectSilenceWindow * 1_000_000_000)
self.preDetectTask = Task { [weak self, lastSeenAt, lastText] in
try? await Task.sleep(nanoseconds: windowNanos)
guard let self else { return }
await self.preDetectSilenceCheck(
lastSeenAt: lastSeenAt,
lastText: lastText,
triggers: triggers,
gateConfig: gateConfig,
config: config)
}
}
private func triggerOnlyPauseCheck(
lastSeenAt: Date?,
lastText: String?,
triggers: [String],
config: RuntimeConfig) async
{
guard !Task.isCancelled else { return }
guard !self.isCapturing else { return }
guard let lastSeenAt, let lastText else { return }
guard self.lastTranscriptAt == lastSeenAt, self.lastTranscript == lastText else { return }
guard self.isTriggerOnly(transcript: lastText, triggers: triggers) else { return }
if let cooldown = self.cooldownUntil, Date() < cooldown {
return
}
self.logger.info("voicewake runtime detected (trigger-only pause)")
await self.beginCapture(command: "", triggerEndTime: nil, config: config)
}
private func textOnlyFallbackMatch(
transcript: String,
triggers: [String],
config: WakeWordGateConfig) -> WakeWordGateMatch?
{
guard let command = VoiceWakeTextUtils.textOnlyCommand(
transcript: transcript,
triggers: triggers,
minCommandLength: config.minCommandLength,
trimWake: Self.trimmedAfterTrigger)
else { return nil }
return WakeWordGateMatch(triggerEndTime: 0, postGap: 0, command: command)
}
private func isTriggerOnly(transcript: String, triggers: [String]) -> Bool {
guard WakeWordGate.matchesTextOnly(text: transcript, triggers: triggers) else { return false }
guard VoiceWakeTextUtils.startsWithTrigger(transcript: transcript, triggers: triggers) else { return false }
return Self.trimmedAfterTrigger(transcript, triggers: triggers).isEmpty
}
private func preDetectSilenceCheck(
lastSeenAt: Date?,
lastText: String?,
triggers: [String],
gateConfig: WakeWordGateConfig,
config: RuntimeConfig) async
{
guard !Task.isCancelled else { return }
guard !self.isCapturing else { return }
guard let lastSeenAt, let lastText else { return }
guard self.lastTranscriptAt == lastSeenAt, self.lastTranscript == lastText else { return }
guard let match = self.textOnlyFallbackMatch(
transcript: lastText,
triggers: triggers,
config: gateConfig)
else { return }
if let cooldown = self.cooldownUntil, Date() < cooldown {
return
}
self.logger.info("voicewake runtime detected (silence fallback) len=\(match.command.count)")
await self.beginCapture(
command: match.command,
triggerEndTime: match.triggerEndTime,
config: config)
}
private func beginCapture(command: String, triggerEndTime: TimeInterval?, config: RuntimeConfig) async {
self.listeningState = .voiceWake
self.isCapturing = true
DiagnosticsFileLog.shared.log(category: "voicewake.runtime", event: "beginCapture")
self.capturedTranscript = command
self.committedTranscript = ""
self.volatileTranscript = command
self.captureStartedAt = Date()
self.cooldownUntil = nil
self.heardBeyondTrigger = !command.isEmpty
self.triggerChimePlayed = false
self.activeTriggerEndTime = triggerEndTime
self.preDetectTask?.cancel()
self.preDetectTask = nil
self.triggerOnlyTask?.cancel()
self.triggerOnlyTask = nil
if config.triggerChime != .none, !self.triggerChimePlayed {
self.triggerChimePlayed = true
await MainActor.run { VoiceWakeChimePlayer.play(config.triggerChime, reason: "voicewake.trigger") }
}
let snapshot = self.committedTranscript + self.volatileTranscript
let attributed = Self.makeAttributed(
committed: self.committedTranscript,
volatile: self.volatileTranscript,
isFinal: false)
self.overlayToken = await MainActor.run {
VoiceSessionCoordinator.shared.startSession(
source: .wakeWord,
text: snapshot,
attributed: attributed,
forwardEnabled: true)
}
// Keep the "ears" boosted for the capture window so the status icon animates while recording.
await MainActor.run { AppStateStore.shared.triggerVoiceEars(ttl: nil) }
self.captureTask?.cancel()
self.captureTask = Task { [weak self] in
guard let self else { return }
await self.monitorCapture(config: config)
}
}
private func monitorCapture(config: RuntimeConfig) async {
let start = self.captureStartedAt ?? Date()
let hardStop = start.addingTimeInterval(self.captureHardStop)
while self.isCapturing {
let now = Date()
if now >= hardStop {
// Hard-stop after a maximum duration so we never leave the recognizer pinned open.
await self.finalizeCapture(config: config)
return
}
let silenceThreshold = self.heardBeyondTrigger ? self.silenceWindow : self.triggerOnlySilenceWindow
if let last = self.lastHeard, now.timeIntervalSince(last) >= silenceThreshold {
await self.finalizeCapture(config: config)
return
}
try? await Task.sleep(nanoseconds: 200_000_000)
}
}
private func finalizeCapture(config: RuntimeConfig) async {
guard self.isCapturing else { return }
self.isCapturing = false
// Disarm trigger matching immediately (before halting recognition) to avoid double-trigger
// races from late callbacks that arrive after isCapturing is cleared.
self.cooldownUntil = Date().addingTimeInterval(self.debounceAfterSend)
self.captureTask?.cancel()
self.captureTask = nil
let finalTranscript = self.capturedTranscript.trimmingCharacters(in: .whitespacesAndNewlines)
DiagnosticsFileLog.shared.log(category: "voicewake.runtime", event: "finalizeCapture", fields: [
"finalLen": "\(finalTranscript.count)",
])
// Stop further recognition events so we don't retrigger immediately with buffered audio.
self.haltRecognitionPipeline()
self.capturedTranscript = ""
self.captureStartedAt = nil
self.lastHeard = nil
self.heardBeyondTrigger = false
self.triggerChimePlayed = false
self.activeTriggerEndTime = nil
self.lastTranscript = nil
self.lastTranscriptAt = nil
self.preDetectTask?.cancel()
self.preDetectTask = nil
self.triggerOnlyTask?.cancel()
self.triggerOnlyTask = nil
await MainActor.run { AppStateStore.shared.stopVoiceEars() }
if let token = self.overlayToken {
await MainActor.run { VoiceSessionCoordinator.shared.updateLevel(token: token, 0) }
}
let delay: TimeInterval = 0.0
let sendChime = finalTranscript.isEmpty ? .none : config.sendChime
if let token = self.overlayToken {
await MainActor.run {
VoiceSessionCoordinator.shared.finalize(
token: token,
text: finalTranscript,
sendChime: sendChime,
autoSendAfter: delay)
}
} else if !finalTranscript.isEmpty {
if sendChime != .none {
await MainActor.run { VoiceWakeChimePlayer.play(sendChime, reason: "voicewake.send") }
}
Task.detached {
await VoiceWakeForwarder.forward(transcript: finalTranscript)
}
}
self.overlayToken = nil
self.scheduleRestartRecognizer()
}
// MARK: - Audio level handling
private func noteAudioLevel(rms: Double) {
guard self.isCapturing else { return }
// Update adaptive noise floor: faster when lower energy (quiet), slower when loud.
let alpha: Double = rms < self.noiseFloorRMS ? 0.08 : 0.01
self.noiseFloorRMS = max(1e-7, self.noiseFloorRMS + (rms - self.noiseFloorRMS) * alpha)
let threshold = max(self.minSpeechRMS, self.noiseFloorRMS * self.speechBoostFactor)
if rms >= threshold {
self.lastHeard = Date()
}
// Normalize against the adaptive threshold so the UI meter stays roughly 0...1 across devices.
let clamped = min(1.0, max(0.0, rms / max(self.minSpeechRMS, threshold)))
if let token = self.overlayToken {
Task { @MainActor in
VoiceSessionCoordinator.shared.updateLevel(token: token, clamped)
}
}
}
private static func rmsLevel(buffer: AVAudioPCMBuffer) -> Double? {
guard let channelData = buffer.floatChannelData?.pointee else { return nil }
let frameCount = Int(buffer.frameLength)
guard frameCount > 0 else { return nil }
var sum: Double = 0
for i in 0..<frameCount {
let sample = Double(channelData[i])
sum += sample * sample
}
return sqrt(sum / Double(frameCount))
}
private func restartRecognizer() {
// Restart the recognizer so we listen for the next trigger with a clean buffer.
let current = self.currentConfig
self.stop(dismissOverlay: false, cancelScheduledRestart: false)
if let current {
Task { await self.start(with: current) }
}
}
private func restartRecognizerIfIdleAndOverlayHidden() async {
if self.isCapturing { return }
self.restartRecognizer()
}
private func scheduleRestartRecognizer(delay: TimeInterval = 0.7) {
self.scheduledRestartTask?.cancel()
self.scheduledRestartTask = Task { [weak self] in
let nanos = UInt64(max(0, delay) * 1_000_000_000)
try? await Task.sleep(nanoseconds: nanos)
guard let self else { return }
await self.consumeScheduledRestart()
await self.restartRecognizerIfIdleAndOverlayHidden()
}
}
private func consumeScheduledRestart() {
self.scheduledRestartTask = nil
}
func applyPushToTalkCooldown() {
self.cooldownUntil = Date().addingTimeInterval(self.debounceAfterSend)
}
func pauseForPushToTalk() {
self.listeningState = .pushToTalk
self.stop(dismissOverlay: false)
}
private func updateHeardBeyondTrigger(withTrimmed trimmed: String) {
if !self.heardBeyondTrigger, !trimmed.isEmpty {
self.heardBeyondTrigger = true
}
}
private static func trimmedAfterTrigger(_ text: String, triggers: [String]) -> String {
let lower = text.lowercased()
for trigger in triggers {
let token = trigger.lowercased().trimmingCharacters(in: .whitespacesAndNewlines)
guard !token.isEmpty, let range = lower.range(of: token) else { continue }
let after = range.upperBound
let trimmed = text[after...].trimmingCharacters(in: .whitespacesAndNewlines)
return String(trimmed)
}
return text
}
private static func commandAfterTrigger(
transcript: String,
segments: [WakeWordSegment],
triggerEndTime: TimeInterval?,
triggers: [String]) -> String
{
guard let triggerEndTime else {
return self.trimmedAfterTrigger(transcript, triggers: triggers)
}
let trimmed = WakeWordGate.commandText(
transcript: transcript,
segments: segments,
triggerEndTime: triggerEndTime)
return trimmed.isEmpty ? self.trimmedAfterTrigger(transcript, triggers: triggers) : trimmed
}
#if DEBUG
static func _testTrimmedAfterTrigger(_ text: String, triggers: [String]) -> String {
self.trimmedAfterTrigger(text, triggers: triggers)
}
static func _testHasContentAfterTrigger(_ text: String, triggers: [String]) -> Bool {
!self.trimmedAfterTrigger(text, triggers: triggers).isEmpty
}
static func _testAttributedColor(isFinal: Bool) -> NSColor {
self.makeAttributed(committed: "sample", volatile: "", isFinal: isFinal)
.attribute(.foregroundColor, at: 0, effectiveRange: nil) as? NSColor ?? .clear
}
#endif
private static func delta(after committed: String, current: String) -> String {
if current.hasPrefix(committed) {
let start = current.index(current.startIndex, offsetBy: committed.count)
return String(current[start...])
}
return current
}
private static func makeAttributed(committed: String, volatile: String, isFinal: Bool) -> NSAttributedString {
let full = NSMutableAttributedString()
let committedAttr: [NSAttributedString.Key: Any] = [
.foregroundColor: NSColor.labelColor,
.font: NSFont.systemFont(ofSize: 13, weight: .regular),
]
full.append(NSAttributedString(string: committed, attributes: committedAttr))
let volatileColor: NSColor = isFinal ? .labelColor : NSColor.tertiaryLabelColor
let volatileAttr: [NSAttributedString.Key: Any] = [
.foregroundColor: volatileColor,
.font: NSFont.systemFont(ofSize: 13, weight: .regular),
]
full.append(NSAttributedString(string: volatile, attributes: volatileAttr))
return full
}
}

View File

@@ -0,0 +1,473 @@
import AVFoundation
import Foundation
import Speech
import SwabbleKit
enum VoiceWakeTestState: Equatable {
case idle
case requesting
case listening
case hearing(String)
case finalizing
case detected(String)
case failed(String)
}
final class VoiceWakeTester {
private let recognizer: SFSpeechRecognizer?
private var audioEngine: AVAudioEngine?
private var recognitionRequest: SFSpeechAudioBufferRecognitionRequest?
private var recognitionTask: SFSpeechRecognitionTask?
private var isStopping = false
private var isFinalizing = false
private var detectionStart: Date?
private var lastHeard: Date?
private var lastLoggedText: String?
private var lastLoggedAt: Date?
private var lastTranscript: String?
private var lastTranscriptAt: Date?
private var silenceTask: Task<Void, Never>?
private var currentTriggers: [String] = []
private var holdingAfterDetect = false
private var detectedText: String?
private let logger = Logger(subsystem: "bot.molt", category: "voicewake")
private let silenceWindow: TimeInterval = 1.0
init(locale: Locale = .current) {
self.recognizer = SFSpeechRecognizer(locale: locale)
}
func start(
triggers: [String],
micID: String?,
localeID: String?,
onUpdate: @escaping @Sendable (VoiceWakeTestState) -> Void) async throws
{
guard self.recognitionTask == nil else { return }
self.isStopping = false
self.isFinalizing = false
self.holdingAfterDetect = false
self.detectedText = nil
self.lastHeard = nil
self.lastLoggedText = nil
self.lastLoggedAt = nil
self.lastTranscript = nil
self.lastTranscriptAt = nil
self.silenceTask?.cancel()
self.silenceTask = nil
self.currentTriggers = triggers
let chosenLocale = localeID.flatMap { Locale(identifier: $0) } ?? Locale.current
let recognizer = SFSpeechRecognizer(locale: chosenLocale)
guard let recognizer, recognizer.isAvailable else {
throw NSError(
domain: "VoiceWakeTester",
code: 1,
userInfo: [NSLocalizedDescriptionKey: "Speech recognition unavailable"])
}
recognizer.defaultTaskHint = .dictation
guard Self.hasPrivacyStrings else {
throw NSError(
domain: "VoiceWakeTester",
code: 3,
userInfo: [
NSLocalizedDescriptionKey: """
Missing mic/speech privacy strings. Rebuild the mac app (scripts/restart-mac.sh) \
to include usage descriptions.
""",
])
}
let granted = try await Self.ensurePermissions()
guard granted else {
throw NSError(
domain: "VoiceWakeTester",
code: 2,
userInfo: [NSLocalizedDescriptionKey: "Microphone or speech permission denied"])
}
self.logInputSelection(preferredMicID: micID)
self.configureSession(preferredMicID: micID)
let engine = AVAudioEngine()
self.audioEngine = engine
self.recognitionRequest = SFSpeechAudioBufferRecognitionRequest()
self.recognitionRequest?.shouldReportPartialResults = true
self.recognitionRequest?.taskHint = .dictation
let request = self.recognitionRequest
let inputNode = engine.inputNode
let format = inputNode.outputFormat(forBus: 0)
guard format.channelCount > 0, format.sampleRate > 0 else {
self.audioEngine = nil
throw NSError(
domain: "VoiceWakeTester",
code: 4,
userInfo: [NSLocalizedDescriptionKey: "No audio input available"])
}
inputNode.removeTap(onBus: 0)
inputNode.installTap(onBus: 0, bufferSize: 2048, format: format) { [weak request] buffer, _ in
request?.append(buffer)
}
engine.prepare()
try engine.start()
DispatchQueue.main.async {
onUpdate(.listening)
}
self.detectionStart = Date()
self.lastHeard = self.detectionStart
guard let request = recognitionRequest else { return }
self.recognitionTask = recognizer.recognitionTask(with: request) { [weak self] result, error in
guard let self, !self.isStopping else { return }
let text = result?.bestTranscription.formattedString ?? ""
let segments = result.map { WakeWordSpeechSegments.from(
transcription: $0.bestTranscription,
transcript: text) } ?? []
let isFinal = result?.isFinal ?? false
let gateConfig = WakeWordGateConfig(triggers: triggers)
var match = WakeWordGate.match(transcript: text, segments: segments, config: gateConfig)
if match == nil, isFinal {
match = self.textOnlyFallbackMatch(
transcript: text,
triggers: triggers,
config: gateConfig)
}
self.maybeLogDebug(
transcript: text,
segments: segments,
triggers: triggers,
match: match,
isFinal: isFinal)
let errorMessage = error?.localizedDescription
Task { [weak self] in
guard let self, !self.isStopping else { return }
await self.handleResult(
match: match,
text: text,
isFinal: isFinal,
errorMessage: errorMessage,
onUpdate: onUpdate)
}
}
}
func stop() {
self.stop(force: true)
}
func finalize(timeout: TimeInterval = 1.5) {
guard self.recognitionTask != nil else {
self.stop(force: true)
return
}
self.isFinalizing = true
self.recognitionRequest?.endAudio()
if let engine = self.audioEngine {
engine.inputNode.removeTap(onBus: 0)
engine.stop()
}
Task { [weak self] in
guard let self else { return }
try? await Task.sleep(nanoseconds: UInt64(timeout * 1_000_000_000))
if !self.isStopping {
self.stop(force: true)
}
}
}
private func stop(force: Bool) {
if force { self.isStopping = true }
self.isFinalizing = false
self.recognitionRequest?.endAudio()
self.recognitionTask?.cancel()
self.recognitionTask = nil
self.recognitionRequest = nil
if let engine = self.audioEngine {
engine.inputNode.removeTap(onBus: 0)
engine.stop()
}
self.audioEngine = nil
self.holdingAfterDetect = false
self.detectedText = nil
self.lastHeard = nil
self.detectionStart = nil
self.lastLoggedText = nil
self.lastLoggedAt = nil
self.lastTranscript = nil
self.lastTranscriptAt = nil
self.silenceTask?.cancel()
self.silenceTask = nil
self.currentTriggers = []
}
private func handleResult(
match: WakeWordGateMatch?,
text: String,
isFinal: Bool,
errorMessage: String?,
onUpdate: @escaping @Sendable (VoiceWakeTestState) -> Void) async
{
if !text.isEmpty {
self.lastHeard = Date()
self.lastTranscript = text
self.lastTranscriptAt = Date()
}
if self.holdingAfterDetect {
return
}
if let match, !match.command.isEmpty {
self.holdingAfterDetect = true
self.detectedText = match.command
self.logger.info("voice wake detected (test) (len=\(match.command.count))")
await MainActor.run { AppStateStore.shared.triggerVoiceEars(ttl: nil) }
self.stop()
await MainActor.run {
AppStateStore.shared.stopVoiceEars()
onUpdate(.detected(match.command))
}
return
}
if !isFinal, !text.isEmpty {
self.scheduleSilenceCheck(
triggers: self.currentTriggers,
onUpdate: onUpdate)
}
if self.isFinalizing {
Task { @MainActor in onUpdate(.finalizing) }
}
if let errorMessage {
self.stop(force: true)
Task { @MainActor in onUpdate(.failed(errorMessage)) }
return
}
if isFinal {
self.stop(force: true)
let state: VoiceWakeTestState = text.isEmpty
? .failed("No speech detected")
: .failed("No trigger heard: “\(text)")
Task { @MainActor in onUpdate(state) }
} else {
let state: VoiceWakeTestState = text.isEmpty ? .listening : .hearing(text)
Task { @MainActor in onUpdate(state) }
}
}
private func maybeLogDebug(
transcript: String,
segments: [WakeWordSegment],
triggers: [String],
match: WakeWordGateMatch?,
isFinal: Bool)
{
guard !transcript.isEmpty else { return }
let level = self.logger.logLevel
guard level == .debug || level == .trace else { return }
if transcript == self.lastLoggedText, !isFinal {
if let last = self.lastLoggedAt, Date().timeIntervalSince(last) < 0.25 {
return
}
}
self.lastLoggedText = transcript
self.lastLoggedAt = Date()
let textOnly = WakeWordGate.matchesTextOnly(text: transcript, triggers: triggers)
let gaps = Self.debugCandidateGaps(triggers: triggers, segments: segments)
let segmentSummary = Self.debugSegments(segments)
let timingCount = segments.count(where: { $0.start > 0 || $0.duration > 0 })
let matchSummary = match.map {
"match=true gap=\(String(format: "%.2f", $0.postGap))s cmdLen=\($0.command.count)"
} ?? "match=false"
self.logger.debug(
"voicewake test transcript='\(transcript, privacy: .private)' textOnly=\(textOnly) " +
"isFinal=\(isFinal) timing=\(timingCount)/\(segments.count) " +
"\(matchSummary) gaps=[\(gaps, privacy: .private)] segments=[\(segmentSummary, privacy: .private)]")
}
private static func debugSegments(_ segments: [WakeWordSegment]) -> String {
segments.map { seg in
let start = String(format: "%.2f", seg.start)
let end = String(format: "%.2f", seg.end)
return "\(seg.text)@\(start)-\(end)"
}.joined(separator: ", ")
}
private static func debugCandidateGaps(triggers: [String], segments: [WakeWordSegment]) -> String {
let tokens = self.normalizeSegments(segments)
guard !tokens.isEmpty else { return "" }
let triggerTokens = self.normalizeTriggers(triggers)
var gaps: [String] = []
for trigger in triggerTokens {
let count = trigger.tokens.count
guard count > 0, tokens.count > count else { continue }
for i in 0...(tokens.count - count - 1) {
let matched = (0..<count).allSatisfy { tokens[i + $0].normalized == trigger.tokens[$0] }
if !matched { continue }
let triggerEnd = tokens[i + count - 1].end
let nextToken = tokens[i + count]
let gap = nextToken.start - triggerEnd
let formatted = String(format: "%.2f", gap)
gaps.append("\(trigger.tokens.joined(separator: " ")):\(formatted)s")
}
}
return gaps.joined(separator: ", ")
}
private struct DebugToken {
let normalized: String
let start: TimeInterval
let end: TimeInterval
}
private struct DebugTriggerTokens {
let tokens: [String]
}
private static func normalizeTriggers(_ triggers: [String]) -> [DebugTriggerTokens] {
var output: [DebugTriggerTokens] = []
for trigger in triggers {
let tokens = trigger
.split(whereSeparator: { $0.isWhitespace })
.map { VoiceWakeTextUtils.normalizeToken(String($0)) }
.filter { !$0.isEmpty }
if tokens.isEmpty { continue }
output.append(DebugTriggerTokens(tokens: tokens))
}
return output
}
private static func normalizeSegments(_ segments: [WakeWordSegment]) -> [DebugToken] {
segments.compactMap { segment in
let normalized = VoiceWakeTextUtils.normalizeToken(segment.text)
guard !normalized.isEmpty else { return nil }
return DebugToken(
normalized: normalized,
start: segment.start,
end: segment.end)
}
}
private func textOnlyFallbackMatch(
transcript: String,
triggers: [String],
config: WakeWordGateConfig) -> WakeWordGateMatch?
{
guard let command = VoiceWakeTextUtils.textOnlyCommand(
transcript: transcript,
triggers: triggers,
minCommandLength: config.minCommandLength,
trimWake: { WakeWordGate.stripWake(text: $0, triggers: $1) })
else { return nil }
return WakeWordGateMatch(triggerEndTime: 0, postGap: 0, command: command)
}
private func holdUntilSilence(onUpdate: @escaping @Sendable (VoiceWakeTestState) -> Void) {
Task { [weak self] in
guard let self else { return }
let detectedAt = Date()
let hardStop = detectedAt.addingTimeInterval(6) // cap overall listen after trigger
while !self.isStopping {
let now = Date()
if now >= hardStop { break }
if let last = self.lastHeard, now.timeIntervalSince(last) >= silenceWindow {
break
}
try? await Task.sleep(nanoseconds: 200_000_000)
}
if !self.isStopping {
self.stop()
await MainActor.run { AppStateStore.shared.stopVoiceEars() }
if let detectedText {
self.logger.info("voice wake hold finished; len=\(detectedText.count)")
Task { @MainActor in onUpdate(.detected(detectedText)) }
}
}
}
}
private func scheduleSilenceCheck(
triggers: [String],
onUpdate: @escaping @Sendable (VoiceWakeTestState) -> Void)
{
self.silenceTask?.cancel()
let lastSeenAt = self.lastTranscriptAt
let lastText = self.lastTranscript
self.silenceTask = Task { [weak self] in
guard let self else { return }
try? await Task.sleep(nanoseconds: UInt64(self.silenceWindow * 1_000_000_000))
guard !Task.isCancelled else { return }
guard !self.isStopping, !self.holdingAfterDetect else { return }
guard let lastSeenAt, let lastText else { return }
guard self.lastTranscriptAt == lastSeenAt, self.lastTranscript == lastText else { return }
guard let match = self.textOnlyFallbackMatch(
transcript: lastText,
triggers: triggers,
config: WakeWordGateConfig(triggers: triggers)) else { return }
self.holdingAfterDetect = true
self.detectedText = match.command
self.logger.info("voice wake detected (test, silence) (len=\(match.command.count))")
await MainActor.run { AppStateStore.shared.triggerVoiceEars(ttl: nil) }
self.stop()
await MainActor.run {
AppStateStore.shared.stopVoiceEars()
onUpdate(.detected(match.command))
}
}
}
private func configureSession(preferredMicID: String?) {
_ = preferredMicID
}
private func logInputSelection(preferredMicID: String?) {
let preferred = (preferredMicID?.isEmpty == false) ? preferredMicID! : "system-default"
self.logger.info(
"voicewake test input preferred=\(preferred, privacy: .public) " +
"\(AudioInputDeviceObserver.defaultInputDeviceSummary(), privacy: .public)")
}
private nonisolated static func ensurePermissions() async throws -> Bool {
let speechStatus = SFSpeechRecognizer.authorizationStatus()
if speechStatus == .notDetermined {
let granted = await withCheckedContinuation { continuation in
SFSpeechRecognizer.requestAuthorization { status in
continuation.resume(returning: status == .authorized)
}
}
guard granted else { return false }
} else if speechStatus != .authorized {
return false
}
let micStatus = AVCaptureDevice.authorizationStatus(for: .audio)
switch micStatus {
case .authorized: return true
case .notDetermined:
return await withCheckedContinuation { continuation in
AVCaptureDevice.requestAccess(for: .audio) { granted in
continuation.resume(returning: granted)
}
}
default:
return false
}
}
private static var hasPrivacyStrings: Bool {
let speech = Bundle.main.object(forInfoDictionaryKey: "NSSpeechRecognitionUsageDescription") as? String
let mic = Bundle.main.object(forInfoDictionaryKey: "NSMicrophoneUsageDescription") as? String
return speech?.isEmpty == false && mic?.isEmpty == false
}
}
extension VoiceWakeTester: @unchecked Sendable {}

View File

@@ -0,0 +1,374 @@
import AppKit
import MoltbotChatUI
import MoltbotKit
import MoltbotProtocol
import Foundation
import OSLog
import QuartzCore
import SwiftUI
private let webChatSwiftLogger = Logger(subsystem: "bot.molt", category: "WebChatSwiftUI")
private enum WebChatSwiftUILayout {
static let windowSize = NSSize(width: 500, height: 840)
static let panelSize = NSSize(width: 480, height: 640)
static let windowMinSize = NSSize(width: 480, height: 360)
static let anchorPadding: CGFloat = 8
}
struct MacGatewayChatTransport: MoltbotChatTransport, Sendable {
func requestHistory(sessionKey: String) async throws -> MoltbotChatHistoryPayload {
try await GatewayConnection.shared.chatHistory(sessionKey: sessionKey)
}
func abortRun(sessionKey: String, runId: String) async throws {
_ = try await GatewayConnection.shared.request(
method: "chat.abort",
params: [
"sessionKey": AnyCodable(sessionKey),
"runId": AnyCodable(runId),
],
timeoutMs: 10000)
}
func listSessions(limit: Int?) async throws -> MoltbotChatSessionsListResponse {
var params: [String: AnyCodable] = [
"includeGlobal": AnyCodable(true),
"includeUnknown": AnyCodable(false),
]
if let limit {
params["limit"] = AnyCodable(limit)
}
let data = try await GatewayConnection.shared.request(
method: "sessions.list",
params: params,
timeoutMs: 15000)
return try JSONDecoder().decode(MoltbotChatSessionsListResponse.self, from: data)
}
func sendMessage(
sessionKey: String,
message: String,
thinking: String,
idempotencyKey: String,
attachments: [MoltbotChatAttachmentPayload]) async throws -> MoltbotChatSendResponse
{
try await GatewayConnection.shared.chatSend(
sessionKey: sessionKey,
message: message,
thinking: thinking,
idempotencyKey: idempotencyKey,
attachments: attachments)
}
func requestHealth(timeoutMs: Int) async throws -> Bool {
try await GatewayConnection.shared.healthOK(timeoutMs: timeoutMs)
}
func events() -> AsyncStream<MoltbotChatTransportEvent> {
AsyncStream { continuation in
let task = Task {
do {
try await GatewayConnection.shared.refresh()
} catch {
webChatSwiftLogger.error("gateway refresh failed \(error.localizedDescription, privacy: .public)")
}
let stream = await GatewayConnection.shared.subscribe()
for await push in stream {
if Task.isCancelled { return }
if let evt = Self.mapPushToTransportEvent(push) {
continuation.yield(evt)
}
}
}
continuation.onTermination = { @Sendable _ in
task.cancel()
}
}
}
static func mapPushToTransportEvent(_ push: GatewayPush) -> MoltbotChatTransportEvent? {
switch push {
case let .snapshot(hello):
let ok = (try? JSONDecoder().decode(
MoltbotGatewayHealthOK.self,
from: JSONEncoder().encode(hello.snapshot.health)))?.ok ?? true
return .health(ok: ok)
case let .event(evt):
switch evt.event {
case "health":
guard let payload = evt.payload else { return nil }
let ok = (try? JSONDecoder().decode(
MoltbotGatewayHealthOK.self,
from: JSONEncoder().encode(payload)))?.ok ?? true
return .health(ok: ok)
case "tick":
return .tick
case "chat":
guard let payload = evt.payload else { return nil }
guard let chat = try? JSONDecoder().decode(
MoltbotChatEventPayload.self,
from: JSONEncoder().encode(payload))
else {
return nil
}
return .chat(chat)
case "agent":
guard let payload = evt.payload else { return nil }
guard let agent = try? JSONDecoder().decode(
MoltbotAgentEventPayload.self,
from: JSONEncoder().encode(payload))
else {
return nil
}
return .agent(agent)
default:
return nil
}
case .seqGap:
return .seqGap
}
}
}
// MARK: - Window controller
@MainActor
final class WebChatSwiftUIWindowController {
private let presentation: WebChatPresentation
private let sessionKey: String
private let hosting: NSHostingController<MoltbotChatView>
private let contentController: NSViewController
private var window: NSWindow?
private var dismissMonitor: Any?
var onClosed: (() -> Void)?
var onVisibilityChanged: ((Bool) -> Void)?
convenience init(sessionKey: String, presentation: WebChatPresentation) {
self.init(sessionKey: sessionKey, presentation: presentation, transport: MacGatewayChatTransport())
}
init(sessionKey: String, presentation: WebChatPresentation, transport: any MoltbotChatTransport) {
self.sessionKey = sessionKey
self.presentation = presentation
let vm = MoltbotChatViewModel(sessionKey: sessionKey, transport: transport)
let accent = Self.color(fromHex: AppStateStore.shared.seamColorHex)
self.hosting = NSHostingController(rootView: MoltbotChatView(
viewModel: vm,
showsSessionSwitcher: true,
userAccent: accent))
self.contentController = Self.makeContentController(for: presentation, hosting: self.hosting)
self.window = Self.makeWindow(for: presentation, contentViewController: self.contentController)
}
deinit {}
var isVisible: Bool {
self.window?.isVisible ?? false
}
func show() {
guard let window else { return }
self.ensureWindowSize()
window.makeKeyAndOrderFront(nil)
NSApp.activate(ignoringOtherApps: true)
self.onVisibilityChanged?(true)
}
func presentAnchored(anchorProvider: () -> NSRect?) {
guard case .panel = self.presentation, let window else { return }
self.installDismissMonitor()
let target = self.reposition(using: anchorProvider)
if !self.isVisible {
let start = target.offsetBy(dx: 0, dy: 8)
window.setFrame(start, display: true)
window.alphaValue = 0
window.makeKeyAndOrderFront(nil)
NSApp.activate(ignoringOtherApps: true)
NSAnimationContext.runAnimationGroup { context in
context.duration = 0.18
context.timingFunction = CAMediaTimingFunction(name: .easeOut)
window.animator().setFrame(target, display: true)
window.animator().alphaValue = 1
}
} else {
window.makeKeyAndOrderFront(nil)
NSApp.activate(ignoringOtherApps: true)
}
self.onVisibilityChanged?(true)
}
func close() {
self.window?.orderOut(nil)
self.onVisibilityChanged?(false)
self.onClosed?()
self.removeDismissMonitor()
}
@discardableResult
private func reposition(using anchorProvider: () -> NSRect?) -> NSRect {
guard let window else { return .zero }
guard let anchor = anchorProvider() else {
let frame = WindowPlacement.topRightFrame(
size: WebChatSwiftUILayout.panelSize,
padding: WebChatSwiftUILayout.anchorPadding)
window.setFrame(frame, display: false)
return frame
}
let screen = NSScreen.screens.first { screen in
screen.frame.contains(anchor.origin) || screen.frame.contains(NSPoint(x: anchor.midX, y: anchor.midY))
} ?? NSScreen.main
let bounds = (screen?.visibleFrame ?? .zero).insetBy(
dx: WebChatSwiftUILayout.anchorPadding,
dy: WebChatSwiftUILayout.anchorPadding)
let frame = WindowPlacement.anchoredBelowFrame(
size: WebChatSwiftUILayout.panelSize,
anchor: anchor,
padding: WebChatSwiftUILayout.anchorPadding,
in: bounds)
window.setFrame(frame, display: false)
return frame
}
private func installDismissMonitor() {
if ProcessInfo.processInfo.isRunningTests { return }
guard self.dismissMonitor == nil, self.window != nil else { return }
self.dismissMonitor = NSEvent.addGlobalMonitorForEvents(
matching: [.leftMouseDown, .rightMouseDown, .otherMouseDown])
{ [weak self] _ in
guard let self, let win = self.window else { return }
let pt = NSEvent.mouseLocation
if !win.frame.contains(pt) {
self.close()
}
}
}
private func removeDismissMonitor() {
if let monitor = self.dismissMonitor {
NSEvent.removeMonitor(monitor)
self.dismissMonitor = nil
}
}
private static func makeWindow(
for presentation: WebChatPresentation,
contentViewController: NSViewController) -> NSWindow
{
switch presentation {
case .window:
let window = NSWindow(
contentRect: NSRect(origin: .zero, size: WebChatSwiftUILayout.windowSize),
styleMask: [.titled, .closable, .resizable, .miniaturizable],
backing: .buffered,
defer: false)
window.title = "Moltbot Chat"
window.contentViewController = contentViewController
window.isReleasedWhenClosed = false
window.titleVisibility = .visible
window.titlebarAppearsTransparent = false
window.backgroundColor = .clear
window.isOpaque = false
window.center()
WindowPlacement.ensureOnScreen(window: window, defaultSize: WebChatSwiftUILayout.windowSize)
window.minSize = WebChatSwiftUILayout.windowMinSize
window.contentView?.wantsLayer = true
window.contentView?.layer?.backgroundColor = NSColor.clear.cgColor
return window
case .panel:
let panel = WebChatPanel(
contentRect: NSRect(origin: .zero, size: WebChatSwiftUILayout.panelSize),
styleMask: [.borderless],
backing: .buffered,
defer: false)
panel.level = .statusBar
panel.hidesOnDeactivate = true
panel.hasShadow = true
panel.isMovable = false
panel.collectionBehavior = [.canJoinAllSpaces, .fullScreenAuxiliary]
panel.titleVisibility = .hidden
panel.titlebarAppearsTransparent = true
panel.backgroundColor = .clear
panel.isOpaque = false
panel.contentViewController = contentViewController
panel.becomesKeyOnlyIfNeeded = true
panel.contentView?.wantsLayer = true
panel.contentView?.layer?.backgroundColor = NSColor.clear.cgColor
panel.setFrame(
WindowPlacement.topRightFrame(
size: WebChatSwiftUILayout.panelSize,
padding: WebChatSwiftUILayout.anchorPadding),
display: false)
return panel
}
}
private static func makeContentController(
for presentation: WebChatPresentation,
hosting: NSHostingController<MoltbotChatView>) -> NSViewController
{
let controller = NSViewController()
let effectView = NSVisualEffectView()
effectView.material = .sidebar
effectView.blendingMode = .behindWindow
effectView.state = .active
effectView.wantsLayer = true
effectView.layer?.cornerCurve = .continuous
let cornerRadius: CGFloat = switch presentation {
case .panel:
16
case .window:
0
}
effectView.layer?.cornerRadius = cornerRadius
effectView.layer?.masksToBounds = true
effectView.translatesAutoresizingMaskIntoConstraints = true
effectView.autoresizingMask = [.width, .height]
let rootView = effectView
hosting.view.translatesAutoresizingMaskIntoConstraints = false
hosting.view.wantsLayer = true
hosting.view.layer?.backgroundColor = NSColor.clear.cgColor
controller.addChild(hosting)
effectView.addSubview(hosting.view)
controller.view = rootView
NSLayoutConstraint.activate([
hosting.view.leadingAnchor.constraint(equalTo: effectView.leadingAnchor),
hosting.view.trailingAnchor.constraint(equalTo: effectView.trailingAnchor),
hosting.view.topAnchor.constraint(equalTo: effectView.topAnchor),
hosting.view.bottomAnchor.constraint(equalTo: effectView.bottomAnchor),
])
return controller
}
private func ensureWindowSize() {
guard case .window = self.presentation, let window else { return }
let current = window.frame.size
let min = WebChatSwiftUILayout.windowMinSize
if current.width < min.width || current.height < min.height {
let frame = WindowPlacement.centeredFrame(size: WebChatSwiftUILayout.windowSize)
window.setFrame(frame, display: false)
}
}
private static func color(fromHex raw: String?) -> Color? {
let trimmed = (raw ?? "").trimmingCharacters(in: .whitespacesAndNewlines)
guard !trimmed.isEmpty else { return nil }
let hex = trimmed.hasPrefix("#") ? String(trimmed.dropFirst()) : trimmed
guard hex.count == 6, let value = Int(hex, radix: 16) else { return nil }
let r = Double((value >> 16) & 0xFF) / 255.0
let g = Double((value >> 8) & 0xFF) / 255.0
let b = Double(value & 0xFF) / 255.0
return Color(red: r, green: g, blue: b)
}
}

View File

@@ -0,0 +1,683 @@
import MoltbotKit
import Foundation
import Network
import Observation
import OSLog
@MainActor
@Observable
public final class GatewayDiscoveryModel {
public struct LocalIdentity: Equatable, Sendable {
public var hostTokens: Set<String>
public var displayTokens: Set<String>
public init(hostTokens: Set<String>, displayTokens: Set<String>) {
self.hostTokens = hostTokens
self.displayTokens = displayTokens
}
}
public struct DiscoveredGateway: Identifiable, Equatable, Sendable {
public var id: String { self.stableID }
public var displayName: String
public var lanHost: String?
public var tailnetDns: String?
public var sshPort: Int
public var gatewayPort: Int?
public var cliPath: String?
public var stableID: String
public var debugID: String
public var isLocal: Bool
public init(
displayName: String,
lanHost: String? = nil,
tailnetDns: String? = nil,
sshPort: Int,
gatewayPort: Int? = nil,
cliPath: String? = nil,
stableID: String,
debugID: String,
isLocal: Bool)
{
self.displayName = displayName
self.lanHost = lanHost
self.tailnetDns = tailnetDns
self.sshPort = sshPort
self.gatewayPort = gatewayPort
self.cliPath = cliPath
self.stableID = stableID
self.debugID = debugID
self.isLocal = isLocal
}
}
public var gateways: [DiscoveredGateway] = []
public var statusText: String = "Idle"
private var browsers: [String: NWBrowser] = [:]
private var resultsByDomain: [String: Set<NWBrowser.Result>] = [:]
private var gatewaysByDomain: [String: [DiscoveredGateway]] = [:]
private var statesByDomain: [String: NWBrowser.State] = [:]
private var localIdentity: LocalIdentity
private let localDisplayName: String?
private let filterLocalGateways: Bool
private var resolvedTXTByID: [String: [String: String]] = [:]
private var pendingTXTResolvers: [String: GatewayTXTResolver] = [:]
private var wideAreaFallbackTask: Task<Void, Never>?
private var wideAreaFallbackGateways: [DiscoveredGateway] = []
private let logger = Logger(subsystem: "bot.molt", category: "gateway-discovery")
public init(
localDisplayName: String? = nil,
filterLocalGateways: Bool = true)
{
self.localDisplayName = localDisplayName
self.filterLocalGateways = filterLocalGateways
self.localIdentity = Self.buildLocalIdentityFast(displayName: localDisplayName)
self.refreshLocalIdentity()
}
public func start() {
if !self.browsers.isEmpty { return }
for domain in MoltbotBonjour.gatewayServiceDomains {
let params = NWParameters.tcp
params.includePeerToPeer = true
let browser = NWBrowser(
for: .bonjour(type: MoltbotBonjour.gatewayServiceType, domain: domain),
using: params)
browser.stateUpdateHandler = { [weak self] state in
Task { @MainActor in
guard let self else { return }
self.statesByDomain[domain] = state
self.updateStatusText()
}
}
browser.browseResultsChangedHandler = { [weak self] results, _ in
Task { @MainActor in
guard let self else { return }
self.resultsByDomain[domain] = results
self.updateGateways(for: domain)
self.recomputeGateways()
}
}
self.browsers[domain] = browser
browser.start(queue: DispatchQueue(label: "bot.molt.macos.gateway-discovery.\(domain)"))
}
self.scheduleWideAreaFallback()
}
public func refreshWideAreaFallbackNow(timeoutSeconds: TimeInterval = 5.0) {
let domain = MoltbotBonjour.wideAreaGatewayServiceDomain
Task.detached(priority: .utility) { [weak self] in
guard let self else { return }
let beacons = WideAreaGatewayDiscovery.discover(timeoutSeconds: timeoutSeconds)
await MainActor.run { [weak self] in
guard let self else { return }
self.wideAreaFallbackGateways = self.mapWideAreaBeacons(beacons, domain: domain)
self.recomputeGateways()
}
}
}
public func stop() {
for browser in self.browsers.values {
browser.cancel()
}
self.browsers = [:]
self.resultsByDomain = [:]
self.gatewaysByDomain = [:]
self.statesByDomain = [:]
self.resolvedTXTByID = [:]
self.pendingTXTResolvers.values.forEach { $0.cancel() }
self.pendingTXTResolvers = [:]
self.wideAreaFallbackTask?.cancel()
self.wideAreaFallbackTask = nil
self.wideAreaFallbackGateways = []
self.gateways = []
self.statusText = "Stopped"
}
private func mapWideAreaBeacons(_ beacons: [WideAreaGatewayBeacon], domain: String) -> [DiscoveredGateway] {
beacons.map { beacon in
let stableID = "wide-area|\(domain)|\(beacon.instanceName)"
let isLocal = Self.isLocalGateway(
lanHost: beacon.lanHost,
tailnetDns: beacon.tailnetDns,
displayName: beacon.displayName,
serviceName: beacon.instanceName,
local: self.localIdentity)
return DiscoveredGateway(
displayName: beacon.displayName,
lanHost: beacon.lanHost,
tailnetDns: beacon.tailnetDns,
sshPort: beacon.sshPort ?? 22,
gatewayPort: beacon.gatewayPort,
cliPath: beacon.cliPath,
stableID: stableID,
debugID: "\(beacon.instanceName)@\(beacon.host):\(beacon.port)",
isLocal: isLocal)
}
}
private func recomputeGateways() {
let primary = self.sortedDeduped(gateways: self.gatewaysByDomain.values.flatMap(\.self))
let primaryFiltered = self.filterLocalGateways ? primary.filter { !$0.isLocal } : primary
if !primaryFiltered.isEmpty {
self.gateways = primaryFiltered
return
}
// Bonjour can return only "local" results for the wide-area domain (or no results at all),
// which makes onboarding look empty even though Tailscale DNS-SD can already see gateways.
guard !self.wideAreaFallbackGateways.isEmpty else {
self.gateways = primaryFiltered
return
}
let combined = self.sortedDeduped(gateways: primary + self.wideAreaFallbackGateways)
self.gateways = self.filterLocalGateways ? combined.filter { !$0.isLocal } : combined
}
private func updateGateways(for domain: String) {
guard let results = self.resultsByDomain[domain] else {
self.gatewaysByDomain[domain] = []
return
}
self.gatewaysByDomain[domain] = results.compactMap { result -> DiscoveredGateway? in
guard case let .service(name, type, resultDomain, _) = result.endpoint else { return nil }
let decodedName = BonjourEscapes.decode(name)
let stableID = GatewayEndpointID.stableID(result.endpoint)
let resolvedTXT = self.resolvedTXTByID[stableID] ?? [:]
let txt = Self.txtDictionary(from: result).merging(
resolvedTXT,
uniquingKeysWith: { _, new in new })
let advertisedName = txt["displayName"]
.map(Self.prettifyInstanceName)
.flatMap { $0.isEmpty ? nil : $0 }
let prettyName =
advertisedName ?? Self.prettifyServiceName(decodedName)
let parsedTXT = Self.parseGatewayTXT(txt)
if parsedTXT.lanHost == nil || parsedTXT.tailnetDns == nil {
self.ensureTXTResolution(
stableID: stableID,
serviceName: name,
type: type,
domain: resultDomain)
}
let isLocal = Self.isLocalGateway(
lanHost: parsedTXT.lanHost,
tailnetDns: parsedTXT.tailnetDns,
displayName: prettyName,
serviceName: decodedName,
local: self.localIdentity)
return DiscoveredGateway(
displayName: prettyName,
lanHost: parsedTXT.lanHost,
tailnetDns: parsedTXT.tailnetDns,
sshPort: parsedTXT.sshPort,
gatewayPort: parsedTXT.gatewayPort,
cliPath: parsedTXT.cliPath,
stableID: stableID,
debugID: GatewayEndpointID.prettyDescription(result.endpoint),
isLocal: isLocal)
}
.sorted { $0.displayName.localizedCaseInsensitiveCompare($1.displayName) == .orderedAscending }
if domain == MoltbotBonjour.wideAreaGatewayServiceDomain,
self.hasUsableWideAreaResults
{
self.wideAreaFallbackGateways = []
}
}
private func scheduleWideAreaFallback() {
let domain = MoltbotBonjour.wideAreaGatewayServiceDomain
if Self.isRunningTests { return }
guard self.wideAreaFallbackTask == nil else { return }
self.wideAreaFallbackTask = Task.detached(priority: .utility) { [weak self] in
guard let self else { return }
var attempt = 0
let startedAt = Date()
while !Task.isCancelled, Date().timeIntervalSince(startedAt) < 35.0 {
let hasResults = await MainActor.run {
self.hasUsableWideAreaResults
}
if hasResults { return }
// Wide-area discovery can be racy (Tailscale not yet up, DNS zone not
// published yet). Retry with a short backoff while onboarding is open.
let beacons = WideAreaGatewayDiscovery.discover(timeoutSeconds: 2.0)
if !beacons.isEmpty {
await MainActor.run { [weak self] in
guard let self else { return }
self.wideAreaFallbackGateways = self.mapWideAreaBeacons(beacons, domain: domain)
self.recomputeGateways()
}
return
}
attempt += 1
let backoff = min(8.0, 0.6 + (Double(attempt) * 0.7))
try? await Task.sleep(nanoseconds: UInt64(backoff * 1_000_000_000))
}
}
}
private var hasUsableWideAreaResults: Bool {
let domain = MoltbotBonjour.wideAreaGatewayServiceDomain
guard let gateways = self.gatewaysByDomain[domain], !gateways.isEmpty else { return false }
if !self.filterLocalGateways { return true }
return gateways.contains(where: { !$0.isLocal })
}
private func sortedDeduped(gateways: [DiscoveredGateway]) -> [DiscoveredGateway] {
var seen = Set<String>()
let deduped = gateways.filter { gateway in
if seen.contains(gateway.stableID) { return false }
seen.insert(gateway.stableID)
return true
}
return deduped.sorted {
$0.displayName.localizedCaseInsensitiveCompare($1.displayName) == .orderedAscending
}
}
private nonisolated static var isRunningTests: Bool {
// Keep discovery background work from running forever during SwiftPM test runs.
if Bundle.allBundles.contains(where: { $0.bundleURL.pathExtension == "xctest" }) { return true }
let env = ProcessInfo.processInfo.environment
return env["XCTestConfigurationFilePath"] != nil
|| env["XCTestBundlePath"] != nil
|| env["XCTestSessionIdentifier"] != nil
}
private func updateGatewaysForAllDomains() {
for domain in self.resultsByDomain.keys {
self.updateGateways(for: domain)
}
}
private func updateStatusText() {
let states = Array(self.statesByDomain.values)
if states.isEmpty {
self.statusText = self.browsers.isEmpty ? "Idle" : "Setup"
return
}
if let failed = states.first(where: { state in
if case .failed = state { return true }
return false
}) {
if case let .failed(err) = failed {
self.statusText = "Failed: \(err)"
return
}
}
if let waiting = states.first(where: { state in
if case .waiting = state { return true }
return false
}) {
if case let .waiting(err) = waiting {
self.statusText = "Waiting: \(err)"
return
}
}
if states.contains(where: { if case .ready = $0 { true } else { false } }) {
self.statusText = "Searching…"
return
}
if states.contains(where: { if case .setup = $0 { true } else { false } }) {
self.statusText = "Setup"
return
}
self.statusText = "Searching…"
}
private static func txtDictionary(from result: NWBrowser.Result) -> [String: String] {
var merged: [String: String] = [:]
if case let .bonjour(txt) = result.metadata {
merged.merge(txt.dictionary, uniquingKeysWith: { _, new in new })
}
if let endpointTxt = result.endpoint.txtRecord?.dictionary {
merged.merge(endpointTxt, uniquingKeysWith: { _, new in new })
}
return merged
}
public struct GatewayTXT: Equatable {
public var lanHost: String?
public var tailnetDns: String?
public var sshPort: Int
public var gatewayPort: Int?
public var cliPath: String?
}
public static func parseGatewayTXT(_ txt: [String: String]) -> GatewayTXT {
var lanHost: String?
var tailnetDns: String?
var sshPort = 22
var gatewayPort: Int?
var cliPath: String?
if let value = txt["lanHost"] {
let trimmed = value.trimmingCharacters(in: .whitespacesAndNewlines)
lanHost = trimmed.isEmpty ? nil : trimmed
}
if let value = txt["tailnetDns"] {
let trimmed = value.trimmingCharacters(in: .whitespacesAndNewlines)
tailnetDns = trimmed.isEmpty ? nil : trimmed
}
if let value = txt["sshPort"],
let parsed = Int(value.trimmingCharacters(in: .whitespacesAndNewlines)),
parsed > 0
{
sshPort = parsed
}
if let value = txt["gatewayPort"],
let parsed = Int(value.trimmingCharacters(in: .whitespacesAndNewlines)),
parsed > 0
{
gatewayPort = parsed
}
if let value = txt["cliPath"] {
let trimmed = value.trimmingCharacters(in: .whitespacesAndNewlines)
cliPath = trimmed.isEmpty ? nil : trimmed
}
return GatewayTXT(
lanHost: lanHost,
tailnetDns: tailnetDns,
sshPort: sshPort,
gatewayPort: gatewayPort,
cliPath: cliPath)
}
public static func buildSSHTarget(user: String, host: String, port: Int) -> String {
var target = "\(user)@\(host)"
if port != 22 {
target += ":\(port)"
}
return target
}
private func ensureTXTResolution(
stableID: String,
serviceName: String,
type: String,
domain: String)
{
guard self.resolvedTXTByID[stableID] == nil else { return }
guard self.pendingTXTResolvers[stableID] == nil else { return }
let resolver = GatewayTXTResolver(
name: serviceName,
type: type,
domain: domain,
logger: self.logger)
{ [weak self] result in
Task { @MainActor in
guard let self else { return }
self.pendingTXTResolvers[stableID] = nil
switch result {
case let .success(txt):
self.resolvedTXTByID[stableID] = txt
self.updateGatewaysForAllDomains()
self.recomputeGateways()
case .failure:
break
}
}
}
self.pendingTXTResolvers[stableID] = resolver
resolver.start()
}
private nonisolated static func prettifyInstanceName(_ decodedName: String) -> String {
let normalized = decodedName.split(whereSeparator: \.isWhitespace).joined(separator: " ")
let stripped = normalized.replacingOccurrences(of: " (Moltbot)", with: "")
.replacingOccurrences(of: #"\s+\(\d+\)$"#, with: "", options: .regularExpression)
return stripped.trimmingCharacters(in: .whitespacesAndNewlines)
}
private nonisolated static func prettifyServiceName(_ decodedName: String) -> String {
let normalized = Self.prettifyInstanceName(decodedName)
var cleaned = normalized.replacingOccurrences(of: #"\s*-?gateway$"#, with: "", options: .regularExpression)
cleaned = cleaned
.replacingOccurrences(of: "_", with: " ")
.replacingOccurrences(of: "-", with: " ")
.replacingOccurrences(of: #"\s+"#, with: " ", options: .regularExpression)
.trimmingCharacters(in: .whitespacesAndNewlines)
if cleaned.isEmpty {
cleaned = normalized
}
let words = cleaned.split(separator: " ")
let titled = words.map { word -> String in
let lower = word.lowercased()
guard let first = lower.first else { return "" }
return String(first).uppercased() + lower.dropFirst()
}.joined(separator: " ")
return titled.isEmpty ? normalized : titled
}
public nonisolated static func isLocalGateway(
lanHost: String?,
tailnetDns: String?,
displayName: String?,
serviceName: String?,
local: LocalIdentity) -> Bool
{
if let host = normalizeHostToken(lanHost),
local.hostTokens.contains(host)
{
return true
}
if let host = normalizeHostToken(tailnetDns),
local.hostTokens.contains(host)
{
return true
}
if let name = normalizeDisplayToken(displayName),
local.displayTokens.contains(name)
{
return true
}
if let serviceHost = normalizeServiceHostToken(serviceName),
local.hostTokens.contains(serviceHost)
{
return true
}
return false
}
private func refreshLocalIdentity() {
let fastIdentity = self.localIdentity
let displayName = self.localDisplayName
Task.detached(priority: .utility) {
let slowIdentity = Self.buildLocalIdentitySlow(displayName: displayName)
let merged = Self.mergeLocalIdentity(fast: fastIdentity, slow: slowIdentity)
await MainActor.run { [weak self] in
guard let self else { return }
guard self.localIdentity != merged else { return }
self.localIdentity = merged
self.recomputeGateways()
}
}
}
private nonisolated static func mergeLocalIdentity(
fast: LocalIdentity,
slow: LocalIdentity) -> LocalIdentity
{
LocalIdentity(
hostTokens: fast.hostTokens.union(slow.hostTokens),
displayTokens: fast.displayTokens.union(slow.displayTokens))
}
private nonisolated static func buildLocalIdentityFast(displayName: String?) -> LocalIdentity {
var hostTokens: Set<String> = []
var displayTokens: Set<String> = []
let hostName = ProcessInfo.processInfo.hostName
if let token = normalizeHostToken(hostName) {
hostTokens.insert(token)
}
if let token = normalizeDisplayToken(displayName) {
displayTokens.insert(token)
}
return LocalIdentity(hostTokens: hostTokens, displayTokens: displayTokens)
}
private nonisolated static func buildLocalIdentitySlow(displayName: String?) -> LocalIdentity {
var hostTokens: Set<String> = []
var displayTokens: Set<String> = []
if let host = Host.current().name,
let token = normalizeHostToken(host)
{
hostTokens.insert(token)
}
if let token = normalizeDisplayToken(displayName) {
displayTokens.insert(token)
}
if let token = normalizeDisplayToken(Host.current().localizedName) {
displayTokens.insert(token)
}
return LocalIdentity(hostTokens: hostTokens, displayTokens: displayTokens)
}
private nonisolated static func normalizeHostToken(_ raw: String?) -> String? {
guard let raw else { return nil }
let trimmed = raw.trimmingCharacters(in: .whitespacesAndNewlines)
if trimmed.isEmpty { return nil }
let lower = trimmed.lowercased()
let strippedTrailingDot = lower.hasSuffix(".")
? String(lower.dropLast())
: lower
let withoutLocal = strippedTrailingDot.hasSuffix(".local")
? String(strippedTrailingDot.dropLast(6))
: strippedTrailingDot
let firstLabel = withoutLocal.split(separator: ".").first.map(String.init)
let token = (firstLabel ?? withoutLocal).trimmingCharacters(in: .whitespacesAndNewlines)
return token.isEmpty ? nil : token
}
private nonisolated static func normalizeDisplayToken(_ raw: String?) -> String? {
guard let raw else { return nil }
let prettified = Self.prettifyInstanceName(raw)
let trimmed = prettified.trimmingCharacters(in: .whitespacesAndNewlines)
if trimmed.isEmpty { return nil }
return trimmed.lowercased()
}
private nonisolated static func normalizeServiceHostToken(_ raw: String?) -> String? {
guard let raw else { return nil }
let prettified = Self.prettifyInstanceName(raw)
let strippedGateway = prettified.replacingOccurrences(
of: #"\s*-?\s*gateway$"#,
with: "",
options: .regularExpression)
return self.normalizeHostToken(strippedGateway)
}
}
final class GatewayTXTResolver: NSObject, NetServiceDelegate {
private let service: NetService
private let completion: (Result<[String: String], Error>) -> Void
private let logger: Logger
private var didFinish = false
init(
name: String,
type: String,
domain: String,
logger: Logger,
completion: @escaping (Result<[String: String], Error>) -> Void)
{
self.service = NetService(domain: domain, type: type, name: name)
self.completion = completion
self.logger = logger
super.init()
self.service.delegate = self
}
func start(timeout: TimeInterval = 2.0) {
self.service.schedule(in: .main, forMode: .common)
self.service.resolve(withTimeout: timeout)
}
func cancel() {
self.finish(result: .failure(GatewayTXTResolverError.cancelled))
}
func netServiceDidResolveAddress(_ sender: NetService) {
let txt = Self.decodeTXT(sender.txtRecordData())
if !txt.isEmpty {
let payload = self.formatTXT(txt)
self.logger.debug(
"discovery: resolved TXT for \(sender.name, privacy: .public): \(payload, privacy: .public)")
}
self.finish(result: .success(txt))
}
func netService(_ sender: NetService, didNotResolve errorDict: [String: NSNumber]) {
self.finish(result: .failure(GatewayTXTResolverError.resolveFailed(errorDict)))
}
private func finish(result: Result<[String: String], Error>) {
guard !self.didFinish else { return }
self.didFinish = true
self.service.stop()
self.service.remove(from: .main, forMode: .common)
self.completion(result)
}
private static func decodeTXT(_ data: Data?) -> [String: String] {
guard let data else { return [:] }
let dict = NetService.dictionary(fromTXTRecord: data)
var out: [String: String] = [:]
out.reserveCapacity(dict.count)
for (key, value) in dict {
if let str = String(data: value, encoding: .utf8) {
out[key] = str
}
}
return out
}
private func formatTXT(_ txt: [String: String]) -> String {
txt.sorted(by: { $0.key < $1.key })
.map { "\($0.key)=\($0.value)" }
.joined(separator: " ")
}
}
enum GatewayTXTResolverError: Error {
case cancelled
case resolveFailed([String: NSNumber])
}

View File

@@ -0,0 +1,61 @@
// swift-tools-version: 6.2
import PackageDescription
let package = Package(
name: "MoltbotKit",
platforms: [
.iOS(.v18),
.macOS(.v15),
],
products: [
.library(name: "MoltbotProtocol", targets: ["MoltbotProtocol"]),
.library(name: "MoltbotKit", targets: ["MoltbotKit"]),
.library(name: "MoltbotChatUI", targets: ["MoltbotChatUI"]),
],
dependencies: [
.package(url: "https://github.com/steipete/ElevenLabsKit", exact: "0.1.0"),
.package(url: "https://github.com/gonzalezreal/textual", exact: "0.3.1"),
],
targets: [
.target(
name: "MoltbotProtocol",
path: "Sources/MoltbotProtocol",
swiftSettings: [
.enableUpcomingFeature("StrictConcurrency"),
]),
.target(
name: "MoltbotKit",
path: "Sources/MoltbotKit",
dependencies: [
"MoltbotProtocol",
.product(name: "ElevenLabsKit", package: "ElevenLabsKit"),
],
resources: [
.process("Resources"),
],
swiftSettings: [
.enableUpcomingFeature("StrictConcurrency"),
]),
.target(
name: "MoltbotChatUI",
path: "Sources/MoltbotChatUI",
dependencies: [
"MoltbotKit",
.product(
name: "Textual",
package: "textual",
condition: .when(platforms: [.macOS, .iOS])),
],
swiftSettings: [
.enableUpcomingFeature("StrictConcurrency"),
]),
.testTarget(
name: "MoltbotKitTests",
dependencies: ["MoltbotKit", "MoltbotChatUI"],
path: "Tests/MoltbotKitTests",
swiftSettings: [
.enableUpcomingFeature("StrictConcurrency"),
.enableExperimentalFeature("SwiftTesting"),
]),
])

View File

@@ -58,7 +58,7 @@ Authoritative list lives in `src/gateway/server.ts` (`METHODS`, `EVENTS`).
- Server handshake + method dispatch: `src/gateway/server.ts` - Server handshake + method dispatch: `src/gateway/server.ts`
- Node client: `src/gateway/client.ts` - Node client: `src/gateway/client.ts`
- Generated JSON Schema: `dist/protocol.schema.json` - Generated JSON Schema: `dist/protocol.schema.json`
- Generated Swift models: `apps/macos/Sources/ClawdbotProtocol/GatewayModels.swift` - Generated Swift models: `apps/macos/Sources/MoltbotProtocol/GatewayModels.swift`
## Current pipeline ## Current pipeline

View File

@@ -115,7 +115,7 @@
"lint:all": "pnpm lint && pnpm lint:swift", "lint:all": "pnpm lint && pnpm lint:swift",
"lint:fix": "pnpm format:fix && oxlint --type-aware --fix src test", "lint:fix": "pnpm format:fix && oxlint --type-aware --fix src test",
"format": "oxfmt --check src test", "format": "oxfmt --check src test",
"format:swift": "swiftformat --lint --config .swiftformat apps/macos/Sources apps/ios/Sources apps/shared/ClawdbotKit/Sources", "format:swift": "swiftformat --lint --config .swiftformat apps/macos/Sources apps/ios/Sources apps/shared/MoltbotKit/Sources",
"format:all": "pnpm format && pnpm format:swift", "format:all": "pnpm format && pnpm format:swift",
"format:fix": "oxfmt --write src test", "format:fix": "oxfmt --write src test",
"test": "node scripts/test-parallel.mjs", "test": "node scripts/test-parallel.mjs",
@@ -141,7 +141,7 @@
"test:install:e2e:anthropic": "CLAWDBOT_E2E_MODELS=anthropic bash scripts/test-install-sh-e2e-docker.sh", "test:install:e2e:anthropic": "CLAWDBOT_E2E_MODELS=anthropic bash scripts/test-install-sh-e2e-docker.sh",
"protocol:gen": "node --import tsx scripts/protocol-gen.ts", "protocol:gen": "node --import tsx scripts/protocol-gen.ts",
"protocol:gen:swift": "node --import tsx scripts/protocol-gen-swift.ts", "protocol:gen:swift": "node --import tsx scripts/protocol-gen-swift.ts",
"protocol:check": "pnpm protocol:gen && pnpm protocol:gen:swift && git diff --exit-code -- dist/protocol.schema.json apps/macos/Sources/ClawdbotProtocol/GatewayModels.swift", "protocol:check": "pnpm protocol:gen && pnpm protocol:gen:swift && git diff --exit-code -- dist/protocol.schema.json apps/macos/Sources/MoltbotProtocol/GatewayModels.swift",
"canvas:a2ui:bundle": "bash scripts/bundle-a2ui.sh", "canvas:a2ui:bundle": "bash scripts/bundle-a2ui.sh",
"check:loc": "node --import tsx scripts/check-ts-max-loc.ts --max 500" "check:loc": "node --import tsx scripts/check-ts-max-loc.ts --max 500"
}, },

View File

@@ -11,7 +11,7 @@ ROOT_DIR="$(cd "$(dirname "${BASH_SOURCE[0]}")/.." && pwd)"
HASH_FILE="$ROOT_DIR/src/canvas-host/a2ui/.bundle.hash" HASH_FILE="$ROOT_DIR/src/canvas-host/a2ui/.bundle.hash"
OUTPUT_FILE="$ROOT_DIR/src/canvas-host/a2ui/a2ui.bundle.js" OUTPUT_FILE="$ROOT_DIR/src/canvas-host/a2ui/a2ui.bundle.js"
A2UI_RENDERER_DIR="$ROOT_DIR/vendor/a2ui/renderers/lit" A2UI_RENDERER_DIR="$ROOT_DIR/vendor/a2ui/renderers/lit"
A2UI_APP_DIR="$ROOT_DIR/apps/shared/ClawdbotKit/Tools/CanvasA2UI" A2UI_APP_DIR="$ROOT_DIR/apps/shared/MoltbotKit/Tools/CanvasA2UI"
# Docker builds exclude vendor/apps via .dockerignore. # Docker builds exclude vendor/apps via .dockerignore.
# In that environment we must keep the prebuilt bundle. # In that environment we must keep the prebuilt bundle.

View File

@@ -24,16 +24,16 @@ const outPaths = [
"apps", "apps",
"macos", "macos",
"Sources", "Sources",
"ClawdbotProtocol", "MoltbotProtocol",
"GatewayModels.swift", "GatewayModels.swift",
), ),
path.join( path.join(
repoRoot, repoRoot,
"apps", "apps",
"shared", "shared",
"ClawdbotKit", "MoltbotKit",
"Sources", "Sources",
"ClawdbotProtocol", "MoltbotProtocol",
"GatewayModels.swift", "GatewayModels.swift",
), ),
]; ];