feat(deeplink): forward agent links via bridge

This commit is contained in:
Peter Steinberger
2025-12-13 01:18:48 +00:00
parent a56daa6c06
commit 378e5acd23
6 changed files with 177 additions and 65 deletions

View File

@@ -10,6 +10,9 @@ struct ClawdisNodeApp: App {
RootCanvas()
.environmentObject(self.appModel)
.environmentObject(self.appModel.voiceWake)
.onOpenURL { url in
Task { await self.appModel.handleDeepLink(url: url) }
}
.onChange(of: self.scenePhase) { _, newValue in
self.appModel.setScenePhase(newValue)
}

View File

@@ -114,6 +114,56 @@ final class NodeAppModel: ObservableObject {
try await self.bridge.sendEvent(event: "voice.transcript", payloadJSON: json)
}
func handleDeepLink(url: URL) async {
guard let route = DeepLinkParser.parse(url) else { return }
switch route {
case let .agent(link):
await self.handleAgentDeepLink(link, originalURL: url)
}
}
private func handleAgentDeepLink(_ link: AgentDeepLink, originalURL: URL) async {
let message = link.message.trimmingCharacters(in: .whitespacesAndNewlines)
guard !message.isEmpty else { return }
if message.count > 20000 {
self.screen.errorText = "Deep link too large (message exceeds 20,000 characters)."
return
}
guard await self.isBridgeConnected() else {
self.screen.errorText = "Bridge not connected (cannot forward deep link)."
return
}
do {
try await self.sendAgentRequest(link: link)
self.screen.errorText = nil
} catch {
self.screen.errorText = "Agent request failed: \(error.localizedDescription)"
}
}
private func sendAgentRequest(link: AgentDeepLink) async throws {
if link.message.trimmingCharacters(in: .whitespacesAndNewlines).isEmpty {
throw NSError(domain: "DeepLink", code: 1, userInfo: [
NSLocalizedDescriptionKey: "invalid agent message",
])
}
// iOS bridge forwards to the gateway; no local auth prompts here.
// (Key-based unattended auth is handled on macOS for clawdis:// links.)
let data = try JSONEncoder().encode(link)
let json = String(decoding: data, as: UTF8.self)
try await self.bridge.sendEvent(event: "agent.request", payloadJSON: json)
}
private func isBridgeConnected() async -> Bool {
if case .connected = await self.bridge.state { return true }
return false
}
private func handleInvoke(_ req: BridgeInvokeRequest) async -> BridgeInvokeResponse {
if req.command.hasPrefix("screen."), self.isBackgrounded {
return BridgeInvokeResponse(

View File

@@ -123,17 +123,24 @@ final class VoiceWakeManager: NSObject, ObservableObject {
self.audioEngine.prepare()
try self.audioEngine.start()
self.recognitionTask = self.speechRecognizer?
.recognitionTask(with: request) { [weak manager = self] result, error in
Task { @MainActor in
manager?.handleRecognitionCallback(result: result, error: error)
}
}
let handler = self.makeRecognitionResultHandler()
self.recognitionTask = self.speechRecognizer?.recognitionTask(with: request, resultHandler: handler)
}
private func handleRecognitionCallback(result: SFSpeechRecognitionResult?, error: Error?) {
if let error {
self.statusText = "Recognizer error: \(error.localizedDescription)"
private nonisolated func makeRecognitionResultHandler() -> @Sendable (SFSpeechRecognitionResult?, Error?) -> Void {
{ [weak self] result, error in
let transcript = result?.bestTranscription.formattedString
let errorText = error?.localizedDescription
Task { @MainActor in
self?.handleRecognitionCallback(transcript: transcript, errorText: errorText)
}
}
}
private func handleRecognitionCallback(transcript: String?, errorText: String?) {
if let errorText {
self.statusText = "Recognizer error: \(errorText)"
self.isListening = false
let shouldRestart = self.isEnabled
@@ -146,8 +153,7 @@ final class VoiceWakeManager: NSObject, ObservableObject {
return
}
guard let result else { return }
let transcript = result.bestTranscription.formattedString
guard let transcript else { return }
guard let cmd = self.extractCommand(from: transcript) else { return }
if cmd == self.lastDispatched { return }
@@ -189,17 +195,21 @@ final class VoiceWakeManager: NSObject, ObservableObject {
}
private nonisolated static func requestMicrophonePermission() async -> Bool {
await withCheckedContinuation(isolation: nil) { cont in
await withCheckedContinuation { cont in
AVAudioApplication.requestRecordPermission { ok in
cont.resume(returning: ok)
Task { @MainActor in
cont.resume(returning: ok)
}
}
}
}
private nonisolated static func requestSpeechPermission() async -> Bool {
await withCheckedContinuation(isolation: nil) { cont in
await withCheckedContinuation { cont in
SFSpeechRecognizer.requestAuthorization { status in
cont.resume(returning: status == .authorized)
Task { @MainActor in
cont.resume(returning: status == .authorized)
}
}
}
}

View File

@@ -145,6 +145,31 @@ actor BridgeServer {
deliver: false,
to: nil,
channel: "last")
case "agent.request":
guard let json = evt.payloadJSON, let data = json.data(using: .utf8) else {
return
}
guard let link = try? JSONDecoder().decode(AgentDeepLink.self, from: data) else {
return
}
let message = link.message.trimmingCharacters(in: .whitespacesAndNewlines)
guard !message.isEmpty else { return }
guard message.count <= 20000 else { return }
let sessionKey = link.sessionKey?.trimmingCharacters(in: .whitespacesAndNewlines).nonEmpty
?? "node-\(nodeId)"
let thinking = link.thinking?.trimmingCharacters(in: .whitespacesAndNewlines).nonEmpty
let to = link.to?.trimmingCharacters(in: .whitespacesAndNewlines).nonEmpty
let channel = link.channel?.trimmingCharacters(in: .whitespacesAndNewlines).nonEmpty
_ = await AgentRPC.shared.send(
text: message,
thinking: thinking,
sessionKey: sessionKey,
deliver: link.deliver,
to: to,
channel: channel ?? "last")
default:
break
}

View File

@@ -1,59 +1,11 @@
import AppKit
import ClawdisNodeKit
import Foundation
import OSLog
import Security
private let deepLinkLogger = Logger(subsystem: "com.steipete.clawdis", category: "DeepLink")
enum DeepLinkRoute: Sendable, Equatable {
case agent(AgentDeepLink)
}
struct AgentDeepLink: Sendable, Equatable {
let message: String
let sessionKey: String?
let thinking: String?
let deliver: Bool
let to: String?
let channel: String?
let timeoutSeconds: Int?
let key: String?
}
enum DeepLinkParser {
static func parse(_ url: URL) -> DeepLinkRoute? {
guard url.scheme?.lowercased() == "clawdis" else { return nil }
guard let host = url.host?.lowercased(), !host.isEmpty else { return nil }
guard let comps = URLComponents(url: url, resolvingAgainstBaseURL: false) else { return nil }
let query = (comps.queryItems ?? []).reduce(into: [String: String]()) { dict, item in
guard let value = item.value else { return }
dict[item.name] = value
}
switch host {
case "agent":
guard let message = query["message"], !message.trimmingCharacters(in: .whitespacesAndNewlines).isEmpty else {
return nil
}
let deliver = (query["deliver"] as NSString?)?.boolValue ?? false
let timeoutSeconds = query["timeoutSeconds"].flatMap { Int($0) }.flatMap { $0 >= 0 ? $0 : nil }
return .agent(
.init(
message: message,
sessionKey: query["sessionKey"],
thinking: query["thinking"],
deliver: deliver,
to: query["to"],
channel: query["channel"],
timeoutSeconds: timeoutSeconds,
key: query["key"]))
default:
return nil
}
}
}
@MainActor
final class DeepLinkHandler {
static let shared = DeepLinkHandler()
@@ -128,7 +80,7 @@ final class DeepLinkHandler {
// MARK: - Auth
static func currentKey() -> String {
Self.expectedKey()
self.expectedKey()
}
private static func expectedKey() -> String {

View File

@@ -0,0 +1,72 @@
import Foundation
public enum DeepLinkRoute: Sendable, Equatable {
case agent(AgentDeepLink)
}
public struct AgentDeepLink: Codable, Sendable, Equatable {
public let message: String
public let sessionKey: String?
public let thinking: String?
public let deliver: Bool
public let to: String?
public let channel: String?
public let timeoutSeconds: Int?
public let key: String?
public init(
message: String,
sessionKey: String?,
thinking: String?,
deliver: Bool,
to: String?,
channel: String?,
timeoutSeconds: Int?,
key: String?)
{
self.message = message
self.sessionKey = sessionKey
self.thinking = thinking
self.deliver = deliver
self.to = to
self.channel = channel
self.timeoutSeconds = timeoutSeconds
self.key = key
}
}
public enum DeepLinkParser {
public static func parse(_ url: URL) -> DeepLinkRoute? {
guard url.scheme?.lowercased() == "clawdis" else { return nil }
guard let host = url.host?.lowercased(), !host.isEmpty else { return nil }
guard let comps = URLComponents(url: url, resolvingAgainstBaseURL: false) else { return nil }
let query = (comps.queryItems ?? []).reduce(into: [String: String]()) { dict, item in
guard let value = item.value else { return }
dict[item.name] = value
}
switch host {
case "agent":
guard let message = query["message"],
!message.trimmingCharacters(in: .whitespacesAndNewlines).isEmpty
else {
return nil
}
let deliver = (query["deliver"] as NSString?)?.boolValue ?? false
let timeoutSeconds = query["timeoutSeconds"].flatMap { Int($0) }.flatMap { $0 >= 0 ? $0 : nil }
return .agent(
.init(
message: message,
sessionKey: query["sessionKey"],
thinking: query["thinking"],
deliver: deliver,
to: query["to"],
channel: query["channel"],
timeoutSeconds: timeoutSeconds,
key: query["key"]))
default:
return nil
}
}
}