@pyai/sdk 0.12.0 → 0.13.0

This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
package/CLI.schema.json CHANGED
@@ -239,7 +239,8 @@
239
239
  "port": "string",
240
240
  "no-browser": "boolean",
241
241
  "name": "string",
242
- "white-label": "boolean"
242
+ "white-label": "boolean",
243
+ "transfers": "boolean"
243
244
  },
244
245
  "example": "pyai dialer --profile production"
245
246
  },
package/README.md CHANGED
@@ -106,12 +106,12 @@ the instant the socket opens and routes server frames to typed callbacks, so you
106
106
  is acked but silently dropped, giving you a connected session with zero turns):
107
107
 
108
108
  <!-- omni-starter:start -->
109
- Requires SDK 0.12.0 for `--template omni`. The runnable starter uses ESM,
109
+ Requires SDK 0.13.0 for `--template omni`. The runnable starter uses ESM,
110
110
  Node 20.19+ with an explicit `ws` transport, or Node 22+. MCP 0.5.1
111
111
  requires Node 22+. The SDK's REST client continues to support Node 18+.
112
112
 
113
113
  ```sh
114
- npm install @pyai/sdk@0.12.0
114
+ npm install @pyai/sdk@0.13.0
115
115
  npx pyai init voice-demo --template omni
116
116
  cd voice-demo
117
117
  npm install
@@ -610,6 +610,28 @@ The standalone app exposes two-line parallel calling in the calling-order menu.
610
610
  Signed native installation and OS-wide meeting capture are separate features.
611
611
  See `softphone-desktop/README.md` for host lifecycle integration and current limits.
612
612
 
613
+ ### Warm and cold transfers
614
+
615
+ Where your PyAI deployment enables transfers, start a single-destination call
616
+ with `transfer_enabled: true`. The server broker exposes optional
617
+ `transfers.create/complete/cancel` methods; retain normal authentication and
618
+ ownership checks when proxying them. The widget then shows **Consult first**
619
+ for warm transfer and **Transfer now** for cold transfer. Warm consultation
620
+ holds the caller; complete only after the target joins, or return to the caller.
621
+ Cold transfer leaves before the target answers.
622
+
623
+ The controller persists the target, retry key and completion/cancellation
624
+ command before sending them. **Recover transfer** reuses that exact intent.
625
+ After successful handoff, **Done** clears only the local wrap-up; the caller's
626
+ conversation, recording and usage continue under the original PyAI call. Closing
627
+ the app after a handoff does not hang up that continuing conversation. A cold
628
+ transfer still connecting cannot be wrapped up as a successful handoff.
629
+
630
+ Open the standalone app with `pyai dialer --transfers` and select **Allow
631
+ transfers** before dialing. This is an operator opt-in; the API independently
632
+ enforces activation and current destination rules. Parallel groups cannot opt
633
+ into transfers. Enabling the local control does not enable the backend feature.
634
+
613
635
  ### Meeting recording
614
636
 
615
637
  Import `MeetingRecorder` and `IndexedDbMeetingRecordingStore` from
@@ -646,7 +668,7 @@ meeting. This web component is not a signed native desktop installer.
646
668
  ### Download and open PyAI Phone
647
669
 
648
670
  ```sh
649
- npm install -g @pyai/sdk@0.12.0
671
+ npm install -g @pyai/sdk@0.13.0
650
672
  pyai login
651
673
  pyai dialer
652
674
  ```
@@ -672,7 +694,7 @@ journal. The app does not dial, buy a number or request a microphone at startup.
672
694
 
673
695
  Manual calling and list/priority queues require a foreground start. Each next
674
696
  call waits for confirmed completion and your Done action. Pause stops the queue
675
- from advancing. Parallel dialing and inbound human ringing are not enabled in
697
+ from advancing. Inbound human ringing is not enabled in
676
698
  this release. A window closure submits an owned stop; reopen the same profile
677
699
  to reconcile a completion that could not be confirmed. Workspace admission,
678
700
  consent, recording and billing policy remain authoritative for every call.
@@ -1,6 +1,6 @@
1
1
  export declare const omniTemplate: Readonly<{
2
- "README.md": "# Runnable Omni starter\n\nRequires Node 20.19+ and ESM. Node 20 uses the included `ws` transport;\nNode 22 also works. The separate MCP server requires Node 22+.\nThis server-side example uses a synthetic read-only office-hours tool. Read\n`PYAI.md` when generating the project through the CLI.\n\n<!-- omni-install:start -->\n```sh\nnpm install @pyai/sdk@0.12.0\nnpx pyai init voice-demo --template omni\ncd voice-demo\nnpm install\n```\n\nInitialization is offline; installing dependencies and running `main.mjs` are\nseparate steps. For an existing checkout of this example, run `npm install`\nin its directory instead.\n<!-- omni-install:end -->\n\nInject `PYAI_API_KEY` through your secret manager or shell environment. Do not\npaste a real key into source, a prompt, or shell history. `.env.example` documents\nthe variable; this program does not load `.env` or saved CLI profiles.\n\nPrepare a 24 kHz PCM16 mono WAV, at most 20 seconds, saying “Please look up the\noffice opening time.” Then run:\n\n```sh\nnode main.mjs caller.wav\n```\n\nRunning the program explicitly opens one bounded Omni session and uploads its\ncaptured reply to Hear. Both consume your key's allowed usage. Automatic retries\nare disabled. No managed number is bought and no phone call is placed.\n\n`session.mjs` waits for configuration and the greeting to drain, sends exactly\none paced input stream (caller PCM or silence), returns the read tool's result,\nand captures output using the rate from `hello.audio_out`. Never run a separate\nsilence timer alongside caller audio. `timing.mjs` is generated from the existing\nOmni evaluation harness's timing utilities.\n\nThe new private `omni-run-*` folder contains `session.wav`, `reply.wav`, and\n`report.json`. The report separates received audio, an office-hours answer\nrecovered by Hear from captured audio, and physical playback (not tested).\nThe office-hours check accepts formatting such as `opens@9a.m`, `9 a.m.` and\n`nine in the morning` in a complete affirmative opening-time sentence. Wrong\ntimes, negation, conflicting or truncated text remain unverified. This is a\nconservative check for the example fact, not a general answer-quality score.\nA synthesis transcript is advisory; it is not evidence that speech arrived.\nThis starter does not negotiate playback receipts. Capture ends after the\nsimulated queue drains and two seconds of quiet, a bounded capture heuristic.\nListen to the saved file when qualifying actual sound quality or completeness.\n\nTo exercise interruption, supply a second WAV saying “Stop speaking now”:\n\n```sh\nnode main.mjs caller.wav interruption.wav\n```\n\nThe second clip starts while reply audio is queued. `onBargeIn` clears that\nsimulated queue, and capture observes at least five seconds after that clip ends.\nAn interrupted answer can correctly fail the full-answer check;\ninspect the separate interruption evidence. This is not a physical speaker test.\nFor a real playback adapter, cancel both queued and currently playing audio.\nReplace the example lookup with authorized application data before product use.\n\nOn a session error, the program exits with JSON containing a recognized machine\n`code` (for example `media_dead` or `invalid_configure`). Unknown server codes\nbecome `unrecognized_server_error`; transport or decoding errors without a\nrecognized transport code become `transport_or_protocol_error`. Raw error text,\ncall IDs and unknown code values are omitted. No automatic retry is attempted;\ninspect the code and your authorized diagnostics before deciding to retry.\n";
3
- "package.json": "{\n \"name\": \"pyai-omni-starter\",\n \"version\": \"0.0.0\",\n \"private\": true,\n \"type\": \"module\",\n \"engines\": {\n \"node\": \">=20.19.0\"\n },\n \"scripts\": {\n \"start\": \"node main.mjs\"\n },\n \"dependencies\": {\n \"@pyai/sdk\": \"0.12.0\",\n \"ws\": \"^8.18.3\"\n }\n}\n";
2
+ "README.md": "# Runnable Omni starter\n\nRequires Node 20.19+ and ESM. Node 20 uses the included `ws` transport;\nNode 22 also works. The separate MCP server requires Node 22+.\nThis server-side example uses a synthetic read-only office-hours tool. Read\n`PYAI.md` when generating the project through the CLI.\n\n<!-- omni-install:start -->\n```sh\nnpm install @pyai/sdk@0.13.0\nnpx pyai init voice-demo --template omni\ncd voice-demo\nnpm install\n```\n\nInitialization is offline; installing dependencies and running `main.mjs` are\nseparate steps. For an existing checkout of this example, run `npm install`\nin its directory instead.\n<!-- omni-install:end -->\n\nInject `PYAI_API_KEY` through your secret manager or shell environment. Do not\npaste a real key into source, a prompt, or shell history. `.env.example` documents\nthe variable; this program does not load `.env` or saved CLI profiles.\n\nPrepare a 24 kHz PCM16 mono WAV, at most 20 seconds, saying “Please look up the\noffice opening time.” Then run:\n\n```sh\nnode main.mjs caller.wav\n```\n\nRunning the program explicitly opens one bounded Omni session and uploads its\ncaptured reply to Hear. Both consume your key's allowed usage. Automatic retries\nare disabled. No managed number is bought and no phone call is placed.\n\n`session.mjs` waits for configuration and the greeting to drain, sends exactly\none paced input stream (caller PCM or silence), returns the read tool's result,\nand captures output using the rate from `hello.audio_out`. Never run a separate\nsilence timer alongside caller audio. `timing.mjs` is generated from the existing\nOmni evaluation harness's timing utilities.\n\nThe new private `omni-run-*` folder contains `session.wav`, `reply.wav`, and\n`report.json`. The report separates received audio, an office-hours answer\nrecovered by Hear from captured audio, and physical playback (not tested).\nThe office-hours check accepts formatting such as `opens@9a.m`, `9 a.m.` and\n`nine in the morning` in a complete affirmative opening-time sentence. Wrong\ntimes, negation, conflicting or truncated text remain unverified. This is a\nconservative check for the example fact, not a general answer-quality score.\nA synthesis transcript is advisory; it is not evidence that speech arrived.\nThis starter does not negotiate playback receipts. Capture ends after the\nsimulated queue drains and two seconds of quiet, a bounded capture heuristic.\nListen to the saved file when qualifying actual sound quality or completeness.\n\nTo exercise interruption, supply a second WAV saying “Stop speaking now”:\n\n```sh\nnode main.mjs caller.wav interruption.wav\n```\n\nThe second clip starts while reply audio is queued. `onBargeIn` clears that\nsimulated queue, and capture observes at least five seconds after that clip ends.\nAn interrupted answer can correctly fail the full-answer check;\ninspect the separate interruption evidence. This is not a physical speaker test.\nFor a real playback adapter, cancel both queued and currently playing audio.\nReplace the example lookup with authorized application data before product use.\n\nOn a session error, the program exits with JSON containing a recognized machine\n`code` (for example `media_dead` or `invalid_configure`). Unknown server codes\nbecome `unrecognized_server_error`; transport or decoding errors without a\nrecognized transport code become `transport_or_protocol_error`. Raw error text,\ncall IDs and unknown code values are omitted. No automatic retry is attempted;\ninspect the code and your authorized diagnostics before deciding to retry.\n";
3
+ "package.json": "{\n \"name\": \"pyai-omni-starter\",\n \"version\": \"0.0.0\",\n \"private\": true,\n \"type\": \"module\",\n \"engines\": {\n \"node\": \">=20.19.0\"\n },\n \"scripts\": {\n \"start\": \"node main.mjs\"\n },\n \"dependencies\": {\n \"@pyai/sdk\": \"0.13.0\",\n \"ws\": \"^8.18.3\"\n }\n}\n";
4
4
  "timing.mjs": "// Monotonic capture timing. Arrival time is distinct from queued playback:\n// a single packet may contain seconds of audio and must finish before settling.\nexport const FRAME_MS = 20;\nexport const REALTIME_GAP_LIMIT_MS = 100;\n// Interruption tests need an observation window after caller transmission,\n// even when an earlier cue/response already finished. These are test bounds,\n// not an inferred engine reply-end or a naturalness threshold.\nexport const INTERRUPTION_OBSERVATION = Object.freeze({\n minimumAfterCallerStreamMs: 5000, quietMs: 2000, maximumAfterCallerStreamMs: 25000,\n});\n\n/** Approximate audible bounds; this is energy detection, not a speech/VAD claim. */\nexport function audibleBounds(pcm, rate, rmsFloor = 160) {\n const window = Math.max(1, Math.round(rate / 100)); // 10 ms windows\n let first = null;\n let last = null;\n for (let off = 0; off < pcm.length; off += window) {\n const end = Math.min(pcm.length, off + window);\n let energy = 0;\n for (let i = off; i < end; i++) energy += pcm[i] * pcm[i];\n if (Math.sqrt(energy / (end - off)) >= rmsFloor) {\n first ??= off;\n last = end;\n }\n }\n return { first, last };\n}\n\nexport function newAudioCapture(started) {\n return {\n started, firstPacketAt: null, lastPacketAt: null,\n firstAudioAt: null, lastAudioAt: null, playbackEndAt: null,\n samples: 0, pcm: [], turnBeginAt: null, eouMs: null,\n latestTurnBeginAt: null, postTurnBeginFirstPacketAt: null,\n postTurnBeginFirstAudioAt: null, lastAssistantTranscriptAt: null, lastCallerTranscriptAt: null,\n callerTranscriptEvents: [], assistantTranscriptEvents: [],\n turnBegins: [], unheardCancellations: [], unheardSupersededTurns: [], tools: [], toolResults: [], kb: null, events: [],\n };\n}\n\nexport function recordTurnBegin(ctx, at, turn, origin = 0) {\n const prior = ctx.turnBegins.at(-1);\n const cancellation = ctx.unheardCancellations.at(-1);\n if (prior && cancellation && !cancellation.admitted\n && cancellation.turn === prior.turn && cancellation.turnBeginIndex === ctx.turnBegins.length - 1\n && Number.isSafeInteger(turn) && turn > prior.turn && at > cancellation.at\n && ctx.samples === 0 && ctx.assistantTranscriptEvents.length === 0\n && ctx.tools.length === 0 && ctx.toolResults.length === 0) {\n cancellation.admitted = true;\n ctx.unheardSupersededTurns.push({ turn: prior.turn, beginAtMs: prior.atMs,\n cancelledAtMs: cancellation.atMs, cancellationEventIndex: cancellation.eventIndex,\n replacementTurn: turn, outputPacketsBeforeReplacement: 0 });\n }\n ctx.turnBeginAt ??= at;\n ctx.latestTurnBeginAt = at;\n // Earlier PCM can be a listening cue. A new reply is still pending even\n // when that cue has already finished playing; retain all of its PCM.\n ctx.postTurnBeginFirstPacketAt = null;\n ctx.postTurnBeginFirstAudioAt = null;\n ctx.turnBegins.push({ atMs: Math.round(at - origin), turn: turn ?? null });\n}\n\nexport function recordUnheardCancellation(ctx, event, at, origin, eventIndex) {\n const prior = ctx.turnBegins.at(-1);\n if (event.event !== \"flush\" || event.reason !== \"turn_merge\"\n || !Number.isSafeInteger(event.cancelled_turn) || event.cancelled_turn <= 0\n || prior?.turn !== event.cancelled_turn || at < ctx.latestTurnBeginAt\n || ctx.samples !== 0 || ctx.assistantTranscriptEvents.length\n || ctx.tools.length || ctx.toolResults.length\n || ctx.unheardCancellations.some(c => c.turnBeginIndex === ctx.turnBegins.length - 1)) return;\n ctx.unheardCancellations.push({ turn: event.cancelled_turn, at,\n atMs: Math.round(at - origin), eventIndex, turnBeginIndex: ctx.turnBegins.length - 1,\n admitted: false });\n}\n\nexport function recordAudio(ctx, pcm, at, rate) {\n if (!(pcm instanceof Int16Array) || !pcm.length) return null;\n const copy = pcm.slice(); // Some transports reuse callback buffers.\n const playbackAt = Math.max(at, ctx.playbackEndAt ?? at);\n const endAt = playbackAt + (copy.length / rate) * 1000;\n const bounds = audibleBounds(copy, rate);\n ctx.firstPacketAt ??= at;\n ctx.lastPacketAt = at;\n ctx.playbackEndAt = endAt;\n const afterTurnBegin = ctx.latestTurnBeginAt != null && at >= ctx.latestTurnBeginAt;\n if (afterTurnBegin) ctx.postTurnBeginFirstPacketAt ??= at;\n if (bounds.first != null) {\n ctx.firstAudioAt ??= playbackAt + (bounds.first / rate) * 1000;\n ctx.lastAudioAt = playbackAt + (bounds.last / rate) * 1000;\n if (afterTurnBegin) ctx.postTurnBeginFirstAudioAt ??= playbackAt + (bounds.first / rate) * 1000;\n }\n ctx.samples += copy.length;\n ctx.pcm.push(copy);\n return { playbackAt, endAt, pcm: copy };\n}\n\n/** Preserve negative gaps: they are overlaps, never missing measurements. */\nexport function captureTiming(ctx, caller, origin = 0) {\n const relative = (at) => at == null ? null : Math.round(at - origin);\n const gap = (at) => at == null || caller.speechOffsetAt == null\n ? null : Math.round(at - caller.speechOffsetAt);\n return {\n ttfbMs: gap(ctx.firstAudioAt),\n anyAudioTtfbMs: gap(ctx.firstAudioAt),\n postTurnBeginTtfbMs: gap(ctx.postTurnBeginFirstAudioAt),\n turnMs: gap(ctx.lastAudioAt),\n packetTtfbMs: gap(ctx.firstPacketAt),\n callerStartMs: relative(caller.startedAt),\n callerSpeechOnsetMs: relative(caller.speechOnsetAt),\n callerSpeechOffsetMs: relative(caller.speechOffsetAt),\n callerStreamEndMs: relative(caller.streamEndAt),\n agentFirstPacketMs: relative(ctx.firstPacketAt),\n agentLastPacketMs: relative(ctx.lastPacketAt),\n agentSpeechOnsetMs: relative(ctx.firstAudioAt),\n agentSpeechOffsetMs: relative(ctx.lastAudioAt),\n agentPlaybackEndMs: relative(ctx.playbackEndAt),\n latestTurnBeginMs: relative(ctx.latestTurnBeginAt),\n postTurnBeginFirstPacketMs: relative(ctx.postTurnBeginFirstPacketAt),\n postTurnBeginSpeechOnsetMs: relative(ctx.postTurnBeginFirstAudioAt),\n lastAssistantTranscriptMs: relative(ctx.lastAssistantTranscriptAt),\n callerMaxFrameGapMs: Math.round(caller.maxFrameGapMs ?? 0),\n method: \"client-monotonic-queued-playback-energy-bounds\",\n };\n}\n\nexport async function streamPcmRealtime(omni, pcm, rate, clock, onFrame = () => {}) {\n const frameSize = Math.max(1, Math.round((rate * FRAME_MS) / 1000));\n const bounds = audibleBounds(pcm, rate);\n const startedAt = clock.now();\n let speechOnsetAt = null;\n let speechOffsetAt = null;\n let previousFrameEnd = startedAt;\n let maxFrameGapMs = 0;\n for (let off = 0; off < pcm.length; off += frameSize) {\n const frame = pcm.subarray(off, Math.min(off + frameSize, pcm.length));\n const at = clock.now();\n maxFrameGapMs = Math.max(maxFrameGapMs, at - previousFrameEnd);\n const durationMs = (frame.length / rate) * 1000;\n omni.sendAudio(frame);\n onFrame(frame, at);\n if (bounds.first != null && bounds.first >= off && bounds.first < off + frame.length) {\n speechOnsetAt = at + ((bounds.first - off) / rate) * 1000;\n }\n if (bounds.last != null && bounds.last > off && bounds.last <= off + frame.length) {\n speechOffsetAt = at + ((bounds.last - off) / rate) * 1000;\n }\n previousFrameEnd = at + durationMs;\n // Pace from the actual send. Never burst late frames to \"catch up\".\n await clock.sleep(Math.max(0, previousFrameEnd - clock.now()));\n }\n return { startedAt, speechOnsetAt, speechOffsetAt,\n streamEndAt: clock.now(), maxFrameGapMs };\n}\n\nexport async function streamSilenceWhile(omni, rate, shouldContinue, clock, onFrame = () => {}) {\n const silence = new Int16Array(Math.max(1, Math.round((rate * FRAME_MS) / 1000)));\n while (shouldContinue()) {\n const at = clock.now();\n omni.sendAudio(silence);\n onFrame(silence, at);\n await clock.sleep(FRAME_MS);\n }\n}\n\nexport async function waitForAgentSettle(getCtx, clock, opts = {}) {\n const started = clock.now();\n const observation = opts.callerStreamEndAt != null;\n if (observation && (!Number.isFinite(opts.callerStreamEndAt)\n || opts.callerStreamEndAt < 0 || opts.callerStreamEndAt > started)) {\n throw new Error(\"Invalid caller observation boundary\");\n }\n const timeoutMs = observation ? INTERRUPTION_OBSERVATION.maximumAfterCallerStreamMs : opts.timeoutMs ?? 25000;\n const settleMs = observation ? INTERRUPTION_OBSERVATION.quietMs : opts.settleMs ?? 2000;\n const deadlineAt = (observation ? opts.callerStreamEndAt : started) + timeoutMs;\n const minimumAt = observation ? opts.callerStreamEndAt + INTERRUPTION_OBSERVATION.minimumAfterCallerStreamMs : -Infinity;\n while (true) {\n const ctx = getCtx();\n const at = clock.now();\n if (opts.isClosed?.()) return { reason: \"closed\", at };\n if (observation && at >= deadlineAt) return { reason: \"timeout\", at };\n const hasResponseAudio = !opts.requireTurnBegin || (ctx.latestTurnBeginAt != null\n && ctx.postTurnBeginFirstAudioAt != null);\n // Synthesis text is progress, not a completion event. It can arrive after\n // early/final audio, so extend the quiet floor without demanding another\n // packet after every advisory. The protocol still has no reply-end marker.\n const activityEnd = Math.max(ctx.playbackEndAt ?? -Infinity,\n ctx.latestTurnBeginAt ?? -Infinity, ctx.lastAssistantTranscriptAt ?? -Infinity,\n observation ? ctx.lastCallerTranscriptAt ?? -Infinity : -Infinity);\n // Do not demand a new post-caller turn_begin: an earlier turn can be\n // coalesced and speak later. This records a bounded quiet observation;\n // downstream response-timing checks can impose a stricter begin boundary.\n if (ctx.firstPacketAt != null && hasResponseAudio && at >= minimumAt && at >= activityEnd + settleMs) {\n return { reason: \"settled\", at };\n }\n if (opts.allowEmpty && ctx.firstPacketAt == null && at >= minimumAt && at - started >= (opts.emptyWaitMs ?? 800)) {\n return { reason: \"empty\", at };\n }\n if (at - started >= timeoutMs) return { reason: \"timeout\", at };\n await clock.sleep(observation ? Math.min(20, deadlineAt - at) : 20);\n }\n}\n\nexport async function withTimeout(promise, ms, message, clock) {\n let timer;\n try {\n return await Promise.race([\n promise,\n new Promise((_, reject) => {\n timer = clock.setTimeout(() => reject(new Error(message)), ms);\n }),\n ]);\n } finally {\n clock.clearTimeout(timer);\n }\n}\n";
5
5
  "main.mjs": "import PyAI from \"@pyai/sdk\";\nimport WebSocket from \"ws\"; // Explicit transport makes this work on Node 20 too.\nimport { readFile, writeFile, mkdtemp } from \"node:fs/promises\";\nimport { join } from \"node:path\";\nimport { runSession } from \"./session.mjs\";\nimport { readWav, writeWav, assessAnswer } from \"./audio.mjs\";\n\nasync function main() {\n const [callerPath, interruptionPath, extra] = process.argv.slice(2);\n if (!callerPath || extra) throw new Error(\"Usage: node main.mjs caller.wav [interruption.wav]\");\n if (!process.env.PYAI_API_KEY) throw new Error(\"Inject PYAI_API_KEY through your environment first\");\n const caller = readWav(await readFile(callerPath));\n const interruption = interruptionPath ? readWav(await readFile(interruptionPath)) : undefined;\n const pyai = new PyAI({ apiKey: process.env.PYAI_API_KEY, maxRetries: 0,\n fetch: (url, options) => fetch(url, { ...options, signal: AbortSignal.timeout(30000) }),\n });\n const { report, audio, replyAudio } = await runSession({ pyai, webSocket: WebSocket, caller, interruption });\n const directory = await mkdtemp(join(process.cwd(), \"omni-run-\"));\n const save = (name, data) => writeFile(join(directory, name), data, { mode: 0o600, flag: \"wx\" });\n await save(\"session.wav\", writeWav(audio, report.output_rate ?? 24000));\n const reply = writeWav(replyAudio, report.output_rate ?? 24000);\n await save(\"reply.wav\", reply);\n // Hear the captured bytes, rather than trusting synthesis-advisory text.\n let capturedTranscript = \"\";\n if (report.reply_audio_bytes) {\n try {\n const heard = await pyai.audio.transcriptions.create({ file: new File([reply], \"reply.wav\", { type: \"audio/wav\" }), model: \"pyai-hear\" });\n capturedTranscript = heard.text;\n } catch { report.hear_error = \"Captured audio saved; Hear verification failed\"; }\n }\n const result = { ...report, captured_transcript: capturedTranscript, evidence: assessAnswer(report, capturedTranscript) };\n await save(\"report.json\", JSON.stringify(result, null, 2) + \"\\n\");\n console.log(JSON.stringify({ directory, ...result }, null, 2));\n if (result.evidence.captured_answer !== \"verified_by_hear\" || (interruption && result.evidence.interruption !== \"simulated_queue_cleared\")) process.exitCode = 1;\n}\n\nmain().catch(error => {\n const message = String(error.message).replaceAll(process.env.PYAI_API_KEY || \"\\0\", \"[REDACTED]\");\n console.error(JSON.stringify({ error: message, ...(error.omni_code ? { code: error.omni_code } : {}) })); process.exitCode = 1;\n});\n";
6
6
  "session.mjs": "import { newAudioCapture, recordAudio, recordTurnBegin, streamPcmRealtime, INTERRUPTION_OBSERVATION } from \"./timing.mjs\";\n\nexport const realClock = {\n now: () => performance.now(),\n sleep: ms => new Promise(resolve => setTimeout(resolve, ms)),\n};\n\n// Only recognized machine codes may leave the example. Error messages, call\n// IDs and arbitrary code values can contain credentials or customer content.\nconst publicErrorCodes = new Set([\n \"invalid_configure\", \"unsupported_tool_transport\", \"tool_configuration_locked\",\n \"media_dead\", \"unauthorized\", \"insufficient_scope\", \"rate_limit_exceeded\",\n \"concurrency_limit_exceeded\", \"daily_cap_exceeded\", \"credit_exhausted\",\n]);\nconst transportErrorCodes = new Set([\"ECONNRESET\", \"ECONNREFUSED\", \"ETIMEDOUT\", \"ENOTFOUND\", \"EAI_AGAIN\"]);\nexport function omniFailure(error) {\n const code = error instanceof Error\n ? (transportErrorCodes.has(error.code) ? error.code : \"transport_or_protocol_error\")\n : (publicErrorCodes.has(error?.code) ? error.code : \"unrecognized_server_error\");\n return Object.assign(new Error(`Omni failed (${code}); no automatic retry was attempted`), { omni_code: code });\n}\n\n/** One bounded, server-side example. Its playback queue is simulated. */\nexport async function runSession({ pyai, webSocket, caller, interruption, clock = realClock, timeoutMs = 60000 }) {\n const rate = 24000, frameSize = 480, silence = new Int16Array(frameSize);\n if (!(caller instanceof Int16Array) || !caller.length || caller.length > rate * 20) throw new Error(\"Provide 1–20 seconds of caller PCM16 at 24 kHz\");\n if (interruption && (!(interruption instanceof Int16Array) || !interruption.length || interruption.length > rate * 20)) throw new Error(\"Invalid interruption PCM\");\n const started = clock.now(), all = newAudioCapture(started), reply = newAudioCapture(started);\n const report = { input_rate: rate, output_rate: null, playback_sink: \"simulated\", tool_calls: 0, tool_executions: 0, duplicate_tool_calls: 0, flushes_after_interruption: 0, cleared_queue_ms: 0, interruption_requested: !!interruption, caller_started_ms: null, interruption_started_ms: null, advisory_transcripts: [], end_reason: \"timeout\" };\n let session, configuredAt = null, closed = false, failure = null, phase = \"greeting\", offset = 0, queueEnd = started, interruptionSent = false, inputEndAt = started;\n const handled = new Set();\n const fail = message => { failure ??= new Error(message); };\n try {\n session = pyai.omni.connect({\n webSocket, rate,\n configure: {\n voice_id: \"stock_amos_en_us\", language: \"en\", greeting: \"Hello. How can I help?\",\n persona: \"When asked about office hours, call lookup_office_hours. Never guess. After the result, say one short sentence giving its opening_time.\",\n tools: [{ name: \"lookup_office_hours\", description: \"Read the synthetic example office opening time.\", side_effect: \"read\", parameters: { type: \"object\", properties: {}, required: [], additionalProperties: false } }],\n },\n onHello: frame => {\n const match = /^pcm16@(8000|24000)$/.exec(frame.audio_out ?? \"\");\n if (!match) return fail(\"Unsupported Omni output format\");\n report.output_rate = Number(match[1]);\n },\n onConfigured: () => { configuredAt ??= clock.now(); },\n onAudio: chunk => {\n if (!report.output_rate) return fail(\"Audio arrived before its output format\");\n const bytes = chunk instanceof ArrayBuffer ? new Uint8Array(chunk) : new Uint8Array(chunk.buffer, chunk.byteOffset, chunk.byteLength);\n if (bytes.byteLength % 2) return fail(\"Invalid PCM16 output\");\n const view = new DataView(bytes.buffer, bytes.byteOffset, bytes.byteLength);\n const pcm = Int16Array.from({ length: bytes.length / 2 }, (_, i) => view.getInt16(i * 2, true));\n const now = clock.now();\n recordAudio(all, pcm, now, report.output_rate);\n queueEnd = Math.max(now, queueEnd) + pcm.length / report.output_rate * 1000;\n if (report.tool_executions) recordAudio(reply, pcm, now, report.output_rate);\n },\n onEvent: frame => {\n if (frame.event === \"turn_begin\") recordTurnBegin(all, clock.now(), frame.turn);\n },\n onTranscript: frame => {\n // An assistant advisory is progress; never treat it as received speech.\n if (report.advisory_transcripts.length < 100) report.advisory_transcripts.push({ role: frame.role, text: frame.text });\n if (frame.role === \"assistant\") all.lastAssistantTranscriptAt = clock.now();\n },\n onToolCall: frame => {\n report.tool_calls++;\n if (handled.has(frame.call_id)) { report.duplicate_tool_calls++; return; }\n handled.add(frame.call_id);\n if (frame.name !== \"lookup_office_hours\") { session.toolResult(frame.call_id, { error: \"Unknown example tool\" }); return; }\n // Replace this read-only fixture with your authorized data lookup.\n report.tool_executions++;\n session.toolResult(frame.call_id, { result: { opening_time: \"9 a.m.\", source: \"synthetic example office\" } });\n },\n onBargeIn: () => {\n const now = clock.now(), queued = Math.max(0, queueEnd - now);\n if (report.interruption_started_ms != null) { report.flushes_after_interruption++; report.cleared_queue_ms += queued; }\n // A real speaker adapter must cancel scheduled and currently playing audio.\n queueEnd = now; all.playbackEndAt = now; reply.playbackEndAt = now;\n },\n onError: error => { failure ??= omniFailure(error); },\n onClose: () => { closed = true; },\n });\n while (clock.now() - started < timeoutMs) {\n if (failure) throw failure;\n if (closed) { report.end_reason = \"closed\"; break; }\n const now = clock.now();\n if (configuredAt == null || !report.output_rate) {\n if (now - started >= 10000) throw new Error(\"Omni connection/configuration timed out\");\n await clock.sleep(20); continue;\n }\n const quietAt = Math.max(queueEnd, all.lastPacketAt ?? now, all.lastAssistantTranscriptAt ?? 0, inputEndAt);\n if (phase === \"greeting\") {\n if (all.firstPacketAt != null && now - configuredAt >= 2000 && now >= quietAt + 800) {\n phase = \"caller\"; report.caller_started_ms = now - started;\n } else if (now - configuredAt > 15000) throw new Error(\"Greeting did not drain within 15 seconds\");\n }\n if (phase === \"reply\" && interruption && !interruptionSent && reply.samples && queueEnd - now >= 200) {\n phase = \"interruption\"; offset = 0; interruptionSent = true; report.interruption_started_ms = now - started;\n }\n const observationEnd = inputEndAt + (interruptionSent ? INTERRUPTION_OBSERVATION.minimumAfterCallerStreamMs : 0);\n if (phase === \"reply\" && reply.samples && now >= quietAt + 2000 && now >= observationEnd && (!interruption || interruptionSent)) {\n report.end_reason = \"quiet_window\"; break;\n }\n // Exactly one producer: each 20 ms slot contains caller PCM OR silence.\n let pcm = silence;\n if (phase === \"caller\" || phase === \"interruption\") {\n const source = phase === \"caller\" ? caller : interruption;\n pcm = source.subarray(offset, offset + frameSize); offset += pcm.length;\n if (offset >= source.length) {\n inputEndAt = now + pcm.length / rate * 1000;\n report[phase === \"caller\" ? \"caller_completed_ms\" : \"interruption_completed_ms\"] = inputEndAt - started;\n phase = \"reply\"; offset = 0;\n }\n }\n await streamPcmRealtime(session, pcm, rate, clock);\n }\n report.audio_bytes = all.samples * 2;\n report.reply_audio_bytes = reply.samples * 2;\n report.elapsed_ms = Math.round(clock.now() - started);\n return { report, audio: all.pcm, replyAudio: reply.pcm };\n } finally { session?.close(); }\n}\n";
@@ -1,7 +1,7 @@
1
1
  // Generated by scripts/generate-omni-starter.mjs; edit examples/omni-starter instead.
2
2
  export const omniTemplate = Object.freeze({
3
- "README.md": "# Runnable Omni starter\n\nRequires Node 20.19+ and ESM. Node 20 uses the included `ws` transport;\nNode 22 also works. The separate MCP server requires Node 22+.\nThis server-side example uses a synthetic read-only office-hours tool. Read\n`PYAI.md` when generating the project through the CLI.\n\n<!-- omni-install:start -->\n```sh\nnpm install @pyai/sdk@0.12.0\nnpx pyai init voice-demo --template omni\ncd voice-demo\nnpm install\n```\n\nInitialization is offline; installing dependencies and running `main.mjs` are\nseparate steps. For an existing checkout of this example, run `npm install`\nin its directory instead.\n<!-- omni-install:end -->\n\nInject `PYAI_API_KEY` through your secret manager or shell environment. Do not\npaste a real key into source, a prompt, or shell history. `.env.example` documents\nthe variable; this program does not load `.env` or saved CLI profiles.\n\nPrepare a 24 kHz PCM16 mono WAV, at most 20 seconds, saying “Please look up the\noffice opening time.” Then run:\n\n```sh\nnode main.mjs caller.wav\n```\n\nRunning the program explicitly opens one bounded Omni session and uploads its\ncaptured reply to Hear. Both consume your key's allowed usage. Automatic retries\nare disabled. No managed number is bought and no phone call is placed.\n\n`session.mjs` waits for configuration and the greeting to drain, sends exactly\none paced input stream (caller PCM or silence), returns the read tool's result,\nand captures output using the rate from `hello.audio_out`. Never run a separate\nsilence timer alongside caller audio. `timing.mjs` is generated from the existing\nOmni evaluation harness's timing utilities.\n\nThe new private `omni-run-*` folder contains `session.wav`, `reply.wav`, and\n`report.json`. The report separates received audio, an office-hours answer\nrecovered by Hear from captured audio, and physical playback (not tested).\nThe office-hours check accepts formatting such as `opens@9a.m`, `9 a.m.` and\n`nine in the morning` in a complete affirmative opening-time sentence. Wrong\ntimes, negation, conflicting or truncated text remain unverified. This is a\nconservative check for the example fact, not a general answer-quality score.\nA synthesis transcript is advisory; it is not evidence that speech arrived.\nThis starter does not negotiate playback receipts. Capture ends after the\nsimulated queue drains and two seconds of quiet, a bounded capture heuristic.\nListen to the saved file when qualifying actual sound quality or completeness.\n\nTo exercise interruption, supply a second WAV saying “Stop speaking now”:\n\n```sh\nnode main.mjs caller.wav interruption.wav\n```\n\nThe second clip starts while reply audio is queued. `onBargeIn` clears that\nsimulated queue, and capture observes at least five seconds after that clip ends.\nAn interrupted answer can correctly fail the full-answer check;\ninspect the separate interruption evidence. This is not a physical speaker test.\nFor a real playback adapter, cancel both queued and currently playing audio.\nReplace the example lookup with authorized application data before product use.\n\nOn a session error, the program exits with JSON containing a recognized machine\n`code` (for example `media_dead` or `invalid_configure`). Unknown server codes\nbecome `unrecognized_server_error`; transport or decoding errors without a\nrecognized transport code become `transport_or_protocol_error`. Raw error text,\ncall IDs and unknown code values are omitted. No automatic retry is attempted;\ninspect the code and your authorized diagnostics before deciding to retry.\n",
4
- "package.json": "{\n \"name\": \"pyai-omni-starter\",\n \"version\": \"0.0.0\",\n \"private\": true,\n \"type\": \"module\",\n \"engines\": {\n \"node\": \">=20.19.0\"\n },\n \"scripts\": {\n \"start\": \"node main.mjs\"\n },\n \"dependencies\": {\n \"@pyai/sdk\": \"0.12.0\",\n \"ws\": \"^8.18.3\"\n }\n}\n",
3
+ "README.md": "# Runnable Omni starter\n\nRequires Node 20.19+ and ESM. Node 20 uses the included `ws` transport;\nNode 22 also works. The separate MCP server requires Node 22+.\nThis server-side example uses a synthetic read-only office-hours tool. Read\n`PYAI.md` when generating the project through the CLI.\n\n<!-- omni-install:start -->\n```sh\nnpm install @pyai/sdk@0.13.0\nnpx pyai init voice-demo --template omni\ncd voice-demo\nnpm install\n```\n\nInitialization is offline; installing dependencies and running `main.mjs` are\nseparate steps. For an existing checkout of this example, run `npm install`\nin its directory instead.\n<!-- omni-install:end -->\n\nInject `PYAI_API_KEY` through your secret manager or shell environment. Do not\npaste a real key into source, a prompt, or shell history. `.env.example` documents\nthe variable; this program does not load `.env` or saved CLI profiles.\n\nPrepare a 24 kHz PCM16 mono WAV, at most 20 seconds, saying “Please look up the\noffice opening time.” Then run:\n\n```sh\nnode main.mjs caller.wav\n```\n\nRunning the program explicitly opens one bounded Omni session and uploads its\ncaptured reply to Hear. Both consume your key's allowed usage. Automatic retries\nare disabled. No managed number is bought and no phone call is placed.\n\n`session.mjs` waits for configuration and the greeting to drain, sends exactly\none paced input stream (caller PCM or silence), returns the read tool's result,\nand captures output using the rate from `hello.audio_out`. Never run a separate\nsilence timer alongside caller audio. `timing.mjs` is generated from the existing\nOmni evaluation harness's timing utilities.\n\nThe new private `omni-run-*` folder contains `session.wav`, `reply.wav`, and\n`report.json`. The report separates received audio, an office-hours answer\nrecovered by Hear from captured audio, and physical playback (not tested).\nThe office-hours check accepts formatting such as `opens@9a.m`, `9 a.m.` and\n`nine in the morning` in a complete affirmative opening-time sentence. Wrong\ntimes, negation, conflicting or truncated text remain unverified. This is a\nconservative check for the example fact, not a general answer-quality score.\nA synthesis transcript is advisory; it is not evidence that speech arrived.\nThis starter does not negotiate playback receipts. Capture ends after the\nsimulated queue drains and two seconds of quiet, a bounded capture heuristic.\nListen to the saved file when qualifying actual sound quality or completeness.\n\nTo exercise interruption, supply a second WAV saying “Stop speaking now”:\n\n```sh\nnode main.mjs caller.wav interruption.wav\n```\n\nThe second clip starts while reply audio is queued. `onBargeIn` clears that\nsimulated queue, and capture observes at least five seconds after that clip ends.\nAn interrupted answer can correctly fail the full-answer check;\ninspect the separate interruption evidence. This is not a physical speaker test.\nFor a real playback adapter, cancel both queued and currently playing audio.\nReplace the example lookup with authorized application data before product use.\n\nOn a session error, the program exits with JSON containing a recognized machine\n`code` (for example `media_dead` or `invalid_configure`). Unknown server codes\nbecome `unrecognized_server_error`; transport or decoding errors without a\nrecognized transport code become `transport_or_protocol_error`. Raw error text,\ncall IDs and unknown code values are omitted. No automatic retry is attempted;\ninspect the code and your authorized diagnostics before deciding to retry.\n",
4
+ "package.json": "{\n \"name\": \"pyai-omni-starter\",\n \"version\": \"0.0.0\",\n \"private\": true,\n \"type\": \"module\",\n \"engines\": {\n \"node\": \">=20.19.0\"\n },\n \"scripts\": {\n \"start\": \"node main.mjs\"\n },\n \"dependencies\": {\n \"@pyai/sdk\": \"0.13.0\",\n \"ws\": \"^8.18.3\"\n }\n}\n",
5
5
  "timing.mjs": "// Monotonic capture timing. Arrival time is distinct from queued playback:\n// a single packet may contain seconds of audio and must finish before settling.\nexport const FRAME_MS = 20;\nexport const REALTIME_GAP_LIMIT_MS = 100;\n// Interruption tests need an observation window after caller transmission,\n// even when an earlier cue/response already finished. These are test bounds,\n// not an inferred engine reply-end or a naturalness threshold.\nexport const INTERRUPTION_OBSERVATION = Object.freeze({\n minimumAfterCallerStreamMs: 5000, quietMs: 2000, maximumAfterCallerStreamMs: 25000,\n});\n\n/** Approximate audible bounds; this is energy detection, not a speech/VAD claim. */\nexport function audibleBounds(pcm, rate, rmsFloor = 160) {\n const window = Math.max(1, Math.round(rate / 100)); // 10 ms windows\n let first = null;\n let last = null;\n for (let off = 0; off < pcm.length; off += window) {\n const end = Math.min(pcm.length, off + window);\n let energy = 0;\n for (let i = off; i < end; i++) energy += pcm[i] * pcm[i];\n if (Math.sqrt(energy / (end - off)) >= rmsFloor) {\n first ??= off;\n last = end;\n }\n }\n return { first, last };\n}\n\nexport function newAudioCapture(started) {\n return {\n started, firstPacketAt: null, lastPacketAt: null,\n firstAudioAt: null, lastAudioAt: null, playbackEndAt: null,\n samples: 0, pcm: [], turnBeginAt: null, eouMs: null,\n latestTurnBeginAt: null, postTurnBeginFirstPacketAt: null,\n postTurnBeginFirstAudioAt: null, lastAssistantTranscriptAt: null, lastCallerTranscriptAt: null,\n callerTranscriptEvents: [], assistantTranscriptEvents: [],\n turnBegins: [], unheardCancellations: [], unheardSupersededTurns: [], tools: [], toolResults: [], kb: null, events: [],\n };\n}\n\nexport function recordTurnBegin(ctx, at, turn, origin = 0) {\n const prior = ctx.turnBegins.at(-1);\n const cancellation = ctx.unheardCancellations.at(-1);\n if (prior && cancellation && !cancellation.admitted\n && cancellation.turn === prior.turn && cancellation.turnBeginIndex === ctx.turnBegins.length - 1\n && Number.isSafeInteger(turn) && turn > prior.turn && at > cancellation.at\n && ctx.samples === 0 && ctx.assistantTranscriptEvents.length === 0\n && ctx.tools.length === 0 && ctx.toolResults.length === 0) {\n cancellation.admitted = true;\n ctx.unheardSupersededTurns.push({ turn: prior.turn, beginAtMs: prior.atMs,\n cancelledAtMs: cancellation.atMs, cancellationEventIndex: cancellation.eventIndex,\n replacementTurn: turn, outputPacketsBeforeReplacement: 0 });\n }\n ctx.turnBeginAt ??= at;\n ctx.latestTurnBeginAt = at;\n // Earlier PCM can be a listening cue. A new reply is still pending even\n // when that cue has already finished playing; retain all of its PCM.\n ctx.postTurnBeginFirstPacketAt = null;\n ctx.postTurnBeginFirstAudioAt = null;\n ctx.turnBegins.push({ atMs: Math.round(at - origin), turn: turn ?? null });\n}\n\nexport function recordUnheardCancellation(ctx, event, at, origin, eventIndex) {\n const prior = ctx.turnBegins.at(-1);\n if (event.event !== \"flush\" || event.reason !== \"turn_merge\"\n || !Number.isSafeInteger(event.cancelled_turn) || event.cancelled_turn <= 0\n || prior?.turn !== event.cancelled_turn || at < ctx.latestTurnBeginAt\n || ctx.samples !== 0 || ctx.assistantTranscriptEvents.length\n || ctx.tools.length || ctx.toolResults.length\n || ctx.unheardCancellations.some(c => c.turnBeginIndex === ctx.turnBegins.length - 1)) return;\n ctx.unheardCancellations.push({ turn: event.cancelled_turn, at,\n atMs: Math.round(at - origin), eventIndex, turnBeginIndex: ctx.turnBegins.length - 1,\n admitted: false });\n}\n\nexport function recordAudio(ctx, pcm, at, rate) {\n if (!(pcm instanceof Int16Array) || !pcm.length) return null;\n const copy = pcm.slice(); // Some transports reuse callback buffers.\n const playbackAt = Math.max(at, ctx.playbackEndAt ?? at);\n const endAt = playbackAt + (copy.length / rate) * 1000;\n const bounds = audibleBounds(copy, rate);\n ctx.firstPacketAt ??= at;\n ctx.lastPacketAt = at;\n ctx.playbackEndAt = endAt;\n const afterTurnBegin = ctx.latestTurnBeginAt != null && at >= ctx.latestTurnBeginAt;\n if (afterTurnBegin) ctx.postTurnBeginFirstPacketAt ??= at;\n if (bounds.first != null) {\n ctx.firstAudioAt ??= playbackAt + (bounds.first / rate) * 1000;\n ctx.lastAudioAt = playbackAt + (bounds.last / rate) * 1000;\n if (afterTurnBegin) ctx.postTurnBeginFirstAudioAt ??= playbackAt + (bounds.first / rate) * 1000;\n }\n ctx.samples += copy.length;\n ctx.pcm.push(copy);\n return { playbackAt, endAt, pcm: copy };\n}\n\n/** Preserve negative gaps: they are overlaps, never missing measurements. */\nexport function captureTiming(ctx, caller, origin = 0) {\n const relative = (at) => at == null ? null : Math.round(at - origin);\n const gap = (at) => at == null || caller.speechOffsetAt == null\n ? null : Math.round(at - caller.speechOffsetAt);\n return {\n ttfbMs: gap(ctx.firstAudioAt),\n anyAudioTtfbMs: gap(ctx.firstAudioAt),\n postTurnBeginTtfbMs: gap(ctx.postTurnBeginFirstAudioAt),\n turnMs: gap(ctx.lastAudioAt),\n packetTtfbMs: gap(ctx.firstPacketAt),\n callerStartMs: relative(caller.startedAt),\n callerSpeechOnsetMs: relative(caller.speechOnsetAt),\n callerSpeechOffsetMs: relative(caller.speechOffsetAt),\n callerStreamEndMs: relative(caller.streamEndAt),\n agentFirstPacketMs: relative(ctx.firstPacketAt),\n agentLastPacketMs: relative(ctx.lastPacketAt),\n agentSpeechOnsetMs: relative(ctx.firstAudioAt),\n agentSpeechOffsetMs: relative(ctx.lastAudioAt),\n agentPlaybackEndMs: relative(ctx.playbackEndAt),\n latestTurnBeginMs: relative(ctx.latestTurnBeginAt),\n postTurnBeginFirstPacketMs: relative(ctx.postTurnBeginFirstPacketAt),\n postTurnBeginSpeechOnsetMs: relative(ctx.postTurnBeginFirstAudioAt),\n lastAssistantTranscriptMs: relative(ctx.lastAssistantTranscriptAt),\n callerMaxFrameGapMs: Math.round(caller.maxFrameGapMs ?? 0),\n method: \"client-monotonic-queued-playback-energy-bounds\",\n };\n}\n\nexport async function streamPcmRealtime(omni, pcm, rate, clock, onFrame = () => {}) {\n const frameSize = Math.max(1, Math.round((rate * FRAME_MS) / 1000));\n const bounds = audibleBounds(pcm, rate);\n const startedAt = clock.now();\n let speechOnsetAt = null;\n let speechOffsetAt = null;\n let previousFrameEnd = startedAt;\n let maxFrameGapMs = 0;\n for (let off = 0; off < pcm.length; off += frameSize) {\n const frame = pcm.subarray(off, Math.min(off + frameSize, pcm.length));\n const at = clock.now();\n maxFrameGapMs = Math.max(maxFrameGapMs, at - previousFrameEnd);\n const durationMs = (frame.length / rate) * 1000;\n omni.sendAudio(frame);\n onFrame(frame, at);\n if (bounds.first != null && bounds.first >= off && bounds.first < off + frame.length) {\n speechOnsetAt = at + ((bounds.first - off) / rate) * 1000;\n }\n if (bounds.last != null && bounds.last > off && bounds.last <= off + frame.length) {\n speechOffsetAt = at + ((bounds.last - off) / rate) * 1000;\n }\n previousFrameEnd = at + durationMs;\n // Pace from the actual send. Never burst late frames to \"catch up\".\n await clock.sleep(Math.max(0, previousFrameEnd - clock.now()));\n }\n return { startedAt, speechOnsetAt, speechOffsetAt,\n streamEndAt: clock.now(), maxFrameGapMs };\n}\n\nexport async function streamSilenceWhile(omni, rate, shouldContinue, clock, onFrame = () => {}) {\n const silence = new Int16Array(Math.max(1, Math.round((rate * FRAME_MS) / 1000)));\n while (shouldContinue()) {\n const at = clock.now();\n omni.sendAudio(silence);\n onFrame(silence, at);\n await clock.sleep(FRAME_MS);\n }\n}\n\nexport async function waitForAgentSettle(getCtx, clock, opts = {}) {\n const started = clock.now();\n const observation = opts.callerStreamEndAt != null;\n if (observation && (!Number.isFinite(opts.callerStreamEndAt)\n || opts.callerStreamEndAt < 0 || opts.callerStreamEndAt > started)) {\n throw new Error(\"Invalid caller observation boundary\");\n }\n const timeoutMs = observation ? INTERRUPTION_OBSERVATION.maximumAfterCallerStreamMs : opts.timeoutMs ?? 25000;\n const settleMs = observation ? INTERRUPTION_OBSERVATION.quietMs : opts.settleMs ?? 2000;\n const deadlineAt = (observation ? opts.callerStreamEndAt : started) + timeoutMs;\n const minimumAt = observation ? opts.callerStreamEndAt + INTERRUPTION_OBSERVATION.minimumAfterCallerStreamMs : -Infinity;\n while (true) {\n const ctx = getCtx();\n const at = clock.now();\n if (opts.isClosed?.()) return { reason: \"closed\", at };\n if (observation && at >= deadlineAt) return { reason: \"timeout\", at };\n const hasResponseAudio = !opts.requireTurnBegin || (ctx.latestTurnBeginAt != null\n && ctx.postTurnBeginFirstAudioAt != null);\n // Synthesis text is progress, not a completion event. It can arrive after\n // early/final audio, so extend the quiet floor without demanding another\n // packet after every advisory. The protocol still has no reply-end marker.\n const activityEnd = Math.max(ctx.playbackEndAt ?? -Infinity,\n ctx.latestTurnBeginAt ?? -Infinity, ctx.lastAssistantTranscriptAt ?? -Infinity,\n observation ? ctx.lastCallerTranscriptAt ?? -Infinity : -Infinity);\n // Do not demand a new post-caller turn_begin: an earlier turn can be\n // coalesced and speak later. This records a bounded quiet observation;\n // downstream response-timing checks can impose a stricter begin boundary.\n if (ctx.firstPacketAt != null && hasResponseAudio && at >= minimumAt && at >= activityEnd + settleMs) {\n return { reason: \"settled\", at };\n }\n if (opts.allowEmpty && ctx.firstPacketAt == null && at >= minimumAt && at - started >= (opts.emptyWaitMs ?? 800)) {\n return { reason: \"empty\", at };\n }\n if (at - started >= timeoutMs) return { reason: \"timeout\", at };\n await clock.sleep(observation ? Math.min(20, deadlineAt - at) : 20);\n }\n}\n\nexport async function withTimeout(promise, ms, message, clock) {\n let timer;\n try {\n return await Promise.race([\n promise,\n new Promise((_, reject) => {\n timer = clock.setTimeout(() => reject(new Error(message)), ms);\n }),\n ]);\n } finally {\n clock.clearTimeout(timer);\n }\n}\n",
6
6
  "main.mjs": "import PyAI from \"@pyai/sdk\";\nimport WebSocket from \"ws\"; // Explicit transport makes this work on Node 20 too.\nimport { readFile, writeFile, mkdtemp } from \"node:fs/promises\";\nimport { join } from \"node:path\";\nimport { runSession } from \"./session.mjs\";\nimport { readWav, writeWav, assessAnswer } from \"./audio.mjs\";\n\nasync function main() {\n const [callerPath, interruptionPath, extra] = process.argv.slice(2);\n if (!callerPath || extra) throw new Error(\"Usage: node main.mjs caller.wav [interruption.wav]\");\n if (!process.env.PYAI_API_KEY) throw new Error(\"Inject PYAI_API_KEY through your environment first\");\n const caller = readWav(await readFile(callerPath));\n const interruption = interruptionPath ? readWav(await readFile(interruptionPath)) : undefined;\n const pyai = new PyAI({ apiKey: process.env.PYAI_API_KEY, maxRetries: 0,\n fetch: (url, options) => fetch(url, { ...options, signal: AbortSignal.timeout(30000) }),\n });\n const { report, audio, replyAudio } = await runSession({ pyai, webSocket: WebSocket, caller, interruption });\n const directory = await mkdtemp(join(process.cwd(), \"omni-run-\"));\n const save = (name, data) => writeFile(join(directory, name), data, { mode: 0o600, flag: \"wx\" });\n await save(\"session.wav\", writeWav(audio, report.output_rate ?? 24000));\n const reply = writeWav(replyAudio, report.output_rate ?? 24000);\n await save(\"reply.wav\", reply);\n // Hear the captured bytes, rather than trusting synthesis-advisory text.\n let capturedTranscript = \"\";\n if (report.reply_audio_bytes) {\n try {\n const heard = await pyai.audio.transcriptions.create({ file: new File([reply], \"reply.wav\", { type: \"audio/wav\" }), model: \"pyai-hear\" });\n capturedTranscript = heard.text;\n } catch { report.hear_error = \"Captured audio saved; Hear verification failed\"; }\n }\n const result = { ...report, captured_transcript: capturedTranscript, evidence: assessAnswer(report, capturedTranscript) };\n await save(\"report.json\", JSON.stringify(result, null, 2) + \"\\n\");\n console.log(JSON.stringify({ directory, ...result }, null, 2));\n if (result.evidence.captured_answer !== \"verified_by_hear\" || (interruption && result.evidence.interruption !== \"simulated_queue_cleared\")) process.exitCode = 1;\n}\n\nmain().catch(error => {\n const message = String(error.message).replaceAll(process.env.PYAI_API_KEY || \"\\0\", \"[REDACTED]\");\n console.error(JSON.stringify({ error: message, ...(error.omni_code ? { code: error.omni_code } : {}) })); process.exitCode = 1;\n});\n",
7
7
  "session.mjs": "import { newAudioCapture, recordAudio, recordTurnBegin, streamPcmRealtime, INTERRUPTION_OBSERVATION } from \"./timing.mjs\";\n\nexport const realClock = {\n now: () => performance.now(),\n sleep: ms => new Promise(resolve => setTimeout(resolve, ms)),\n};\n\n// Only recognized machine codes may leave the example. Error messages, call\n// IDs and arbitrary code values can contain credentials or customer content.\nconst publicErrorCodes = new Set([\n \"invalid_configure\", \"unsupported_tool_transport\", \"tool_configuration_locked\",\n \"media_dead\", \"unauthorized\", \"insufficient_scope\", \"rate_limit_exceeded\",\n \"concurrency_limit_exceeded\", \"daily_cap_exceeded\", \"credit_exhausted\",\n]);\nconst transportErrorCodes = new Set([\"ECONNRESET\", \"ECONNREFUSED\", \"ETIMEDOUT\", \"ENOTFOUND\", \"EAI_AGAIN\"]);\nexport function omniFailure(error) {\n const code = error instanceof Error\n ? (transportErrorCodes.has(error.code) ? error.code : \"transport_or_protocol_error\")\n : (publicErrorCodes.has(error?.code) ? error.code : \"unrecognized_server_error\");\n return Object.assign(new Error(`Omni failed (${code}); no automatic retry was attempted`), { omni_code: code });\n}\n\n/** One bounded, server-side example. Its playback queue is simulated. */\nexport async function runSession({ pyai, webSocket, caller, interruption, clock = realClock, timeoutMs = 60000 }) {\n const rate = 24000, frameSize = 480, silence = new Int16Array(frameSize);\n if (!(caller instanceof Int16Array) || !caller.length || caller.length > rate * 20) throw new Error(\"Provide 1–20 seconds of caller PCM16 at 24 kHz\");\n if (interruption && (!(interruption instanceof Int16Array) || !interruption.length || interruption.length > rate * 20)) throw new Error(\"Invalid interruption PCM\");\n const started = clock.now(), all = newAudioCapture(started), reply = newAudioCapture(started);\n const report = { input_rate: rate, output_rate: null, playback_sink: \"simulated\", tool_calls: 0, tool_executions: 0, duplicate_tool_calls: 0, flushes_after_interruption: 0, cleared_queue_ms: 0, interruption_requested: !!interruption, caller_started_ms: null, interruption_started_ms: null, advisory_transcripts: [], end_reason: \"timeout\" };\n let session, configuredAt = null, closed = false, failure = null, phase = \"greeting\", offset = 0, queueEnd = started, interruptionSent = false, inputEndAt = started;\n const handled = new Set();\n const fail = message => { failure ??= new Error(message); };\n try {\n session = pyai.omni.connect({\n webSocket, rate,\n configure: {\n voice_id: \"stock_amos_en_us\", language: \"en\", greeting: \"Hello. How can I help?\",\n persona: \"When asked about office hours, call lookup_office_hours. Never guess. After the result, say one short sentence giving its opening_time.\",\n tools: [{ name: \"lookup_office_hours\", description: \"Read the synthetic example office opening time.\", side_effect: \"read\", parameters: { type: \"object\", properties: {}, required: [], additionalProperties: false } }],\n },\n onHello: frame => {\n const match = /^pcm16@(8000|24000)$/.exec(frame.audio_out ?? \"\");\n if (!match) return fail(\"Unsupported Omni output format\");\n report.output_rate = Number(match[1]);\n },\n onConfigured: () => { configuredAt ??= clock.now(); },\n onAudio: chunk => {\n if (!report.output_rate) return fail(\"Audio arrived before its output format\");\n const bytes = chunk instanceof ArrayBuffer ? new Uint8Array(chunk) : new Uint8Array(chunk.buffer, chunk.byteOffset, chunk.byteLength);\n if (bytes.byteLength % 2) return fail(\"Invalid PCM16 output\");\n const view = new DataView(bytes.buffer, bytes.byteOffset, bytes.byteLength);\n const pcm = Int16Array.from({ length: bytes.length / 2 }, (_, i) => view.getInt16(i * 2, true));\n const now = clock.now();\n recordAudio(all, pcm, now, report.output_rate);\n queueEnd = Math.max(now, queueEnd) + pcm.length / report.output_rate * 1000;\n if (report.tool_executions) recordAudio(reply, pcm, now, report.output_rate);\n },\n onEvent: frame => {\n if (frame.event === \"turn_begin\") recordTurnBegin(all, clock.now(), frame.turn);\n },\n onTranscript: frame => {\n // An assistant advisory is progress; never treat it as received speech.\n if (report.advisory_transcripts.length < 100) report.advisory_transcripts.push({ role: frame.role, text: frame.text });\n if (frame.role === \"assistant\") all.lastAssistantTranscriptAt = clock.now();\n },\n onToolCall: frame => {\n report.tool_calls++;\n if (handled.has(frame.call_id)) { report.duplicate_tool_calls++; return; }\n handled.add(frame.call_id);\n if (frame.name !== \"lookup_office_hours\") { session.toolResult(frame.call_id, { error: \"Unknown example tool\" }); return; }\n // Replace this read-only fixture with your authorized data lookup.\n report.tool_executions++;\n session.toolResult(frame.call_id, { result: { opening_time: \"9 a.m.\", source: \"synthetic example office\" } });\n },\n onBargeIn: () => {\n const now = clock.now(), queued = Math.max(0, queueEnd - now);\n if (report.interruption_started_ms != null) { report.flushes_after_interruption++; report.cleared_queue_ms += queued; }\n // A real speaker adapter must cancel scheduled and currently playing audio.\n queueEnd = now; all.playbackEndAt = now; reply.playbackEndAt = now;\n },\n onError: error => { failure ??= omniFailure(error); },\n onClose: () => { closed = true; },\n });\n while (clock.now() - started < timeoutMs) {\n if (failure) throw failure;\n if (closed) { report.end_reason = \"closed\"; break; }\n const now = clock.now();\n if (configuredAt == null || !report.output_rate) {\n if (now - started >= 10000) throw new Error(\"Omni connection/configuration timed out\");\n await clock.sleep(20); continue;\n }\n const quietAt = Math.max(queueEnd, all.lastPacketAt ?? now, all.lastAssistantTranscriptAt ?? 0, inputEndAt);\n if (phase === \"greeting\") {\n if (all.firstPacketAt != null && now - configuredAt >= 2000 && now >= quietAt + 800) {\n phase = \"caller\"; report.caller_started_ms = now - started;\n } else if (now - configuredAt > 15000) throw new Error(\"Greeting did not drain within 15 seconds\");\n }\n if (phase === \"reply\" && interruption && !interruptionSent && reply.samples && queueEnd - now >= 200) {\n phase = \"interruption\"; offset = 0; interruptionSent = true; report.interruption_started_ms = now - started;\n }\n const observationEnd = inputEndAt + (interruptionSent ? INTERRUPTION_OBSERVATION.minimumAfterCallerStreamMs : 0);\n if (phase === \"reply\" && reply.samples && now >= quietAt + 2000 && now >= observationEnd && (!interruption || interruptionSent)) {\n report.end_reason = \"quiet_window\"; break;\n }\n // Exactly one producer: each 20 ms slot contains caller PCM OR silence.\n let pcm = silence;\n if (phase === \"caller\" || phase === \"interruption\") {\n const source = phase === \"caller\" ? caller : interruption;\n pcm = source.subarray(offset, offset + frameSize); offset += pcm.length;\n if (offset >= source.length) {\n inputEndAt = now + pcm.length / rate * 1000;\n report[phase === \"caller\" ? \"caller_completed_ms\" : \"interruption_completed_ms\"] = inputEndAt - started;\n phase = \"reply\"; offset = 0;\n }\n }\n await streamPcmRealtime(session, pcm, rate, clock);\n }\n report.audio_bytes = all.samples * 2;\n report.reply_audio_bytes = reply.samples * 2;\n report.elapsed_ms = Math.round(clock.now() - started);\n return { report, audio: all.pcm, replyAudio: reply.pcm };\n } finally { session?.close(); }\n}\n",
package/dist/cli.js CHANGED
@@ -42,7 +42,7 @@ const commands = [
42
42
  { command: "init", description: "Create a new project with API context and request examples offline", args: ["directory"], flags: { template: "string" }, example: "pyai init voice-demo --template typescript" },
43
43
  { command: "smoke", description: "Check models, voices, and speech synthesis", flags: { "tolerate-upstream": "boolean" } },
44
44
  { command: "doctor", description: "Diagnose key, catalogs, and a Speak to Hear round-trip" },
45
- { command: "dialer", description: "Open the private desktop calling and meeting-recording app", flags: { port: "string", "no-browser": "boolean", name: "string", "white-label": "boolean" }, example: "pyai dialer --profile production" },
45
+ { command: "dialer", description: "Open the private desktop calling and meeting-recording app", flags: { port: "string", "no-browser": "boolean", name: "string", "white-label": "boolean", transfers: "boolean" }, example: "pyai dialer --profile production" },
46
46
  ...routes.map((r) => ({ command: r.command, description: r.description, args: r.id ? ["id"] : [], flags: {
47
47
  ...(r.body ? bodyFlags : {}), ...(r.binary ? outputFlags : {}), ...(r.wait ? waitFlags : {}),
48
48
  ...(r.query ?? {}), ...(r.confirm ? { confirm: "boolean" } : {}),
@@ -512,7 +512,7 @@ async function main() {
512
512
  const { startDialerApp } = await import("./dialer/appServer.js");
513
513
  const fingerprint = createHash("sha256").update(config.baseURL).update("\0").update(config.apiKey).digest("hex");
514
514
  const sdk = new PyAI({ apiKey: config.apiKey, baseURL: config.baseURL, fetch: (url, init) => fetch(url, { ...init, redirect: "error", signal: init?.signal ?? AbortSignal.timeout(15000) }) });
515
- const app = await startDialerApp({ sdk, userId: `desktop_${fingerprint.slice(0, 40)}`, scope: `desktop:${fingerprint}`, port, name: value(flags, "name"), whiteLabel: flags["white-label"] === true });
515
+ const app = await startDialerApp({ sdk, userId: `desktop_${fingerprint.slice(0, 40)}`, scope: `desktop:${fingerprint}`, port, name: value(flags, "name"), whiteLabel: flags["white-label"] === true, allowTransfers: flags.transfers === true });
516
516
  output({ url: app.url, local_only: true }, flags, `PyAI Phone is ready at ${app.url}\nKeep this window open. Press Ctrl+C to stop the app and end its owned calls.`);
517
517
  if (flags["no-browser"] || !(await openBrowser(app.launchUrl)))
518
518
  process.stderr.write(`Private one-use launch link (valid for 60 seconds): ${app.launchUrl}\n`);
@@ -6,6 +6,8 @@ export interface DialerAppOptions {
6
6
  scope: string;
7
7
  name?: string;
8
8
  whiteLabel?: boolean;
9
+ /** Operator opt-in; the API independently enforces transfer activation. */
10
+ allowTransfers?: boolean;
9
11
  port?: number;
10
12
  /** Explicit asset injection for packaging/tests, never selected by HTTP. */
11
13
  assets?: {
@@ -7,6 +7,7 @@ import { createRequire } from "node:module";
7
7
  import { dirname, join } from "node:path";
8
8
  import { Readable } from "node:stream";
9
9
  import { createServerDialerBroker } from "./serverBroker.js";
10
+ import { dialerCallHandedOff } from "./managedDialer.js";
10
11
  const terminal = (call) => ["completed", "failed", "canceled"].includes(call.status);
11
12
  const secretMatches = (input, expected) => typeof input === "string" && /^[A-Za-z0-9_-]+$/.test(input) && input.length === expected.length && timingSafeEqual(Buffer.from(input), Buffer.from(expected));
12
13
  const browserModules = new Set(["managedDialer", "dialingQueue", "indexedDbJournal", "voiceSdkAudio", "dialerElement", "meetingRecorder", "meetingRecordingStore", "meetingRecorderElement"]);
@@ -61,7 +62,7 @@ export async function startDialerApp(options) {
61
62
  let bootstrapUsed = false;
62
63
  let origin = "", closing = false;
63
64
  const calls = new Set();
64
- const remember = (call) => { if (terminal(call))
65
+ const remember = (call) => { if (terminal(call) || dialerCallHandedOff(call))
65
66
  calls.delete(call.id);
66
67
  else
67
68
  calls.add(call.id); return call; };
@@ -99,19 +100,35 @@ export async function startDialerApp(options) {
99
100
  }
100
101
  if (url.pathname === "/api/config" && method === "GET") {
101
102
  const owned = await numbers();
102
- return json(response, 200, { name: options.name ?? (options.whiteLabel ? "Phone" : "PyAI Phone"), white_label: options.whiteLabel === true, scope: options.scope, user_id: options.userId, nonce,
103
+ return json(response, 200, { name: options.name ?? (options.whiteLabel ? "Phone" : "PyAI Phone"), white_label: options.whiteLabel === true, transfers_enabled: options.allowTransfers === true, scope: options.scope, user_id: options.userId, nonce,
103
104
  numbers: owned.map(number => ({ id: number.id, phone_number: number.phone_number, country: number.country })) });
104
105
  }
105
106
  if (url.pathname === "/api/sessions" && method === "POST") {
106
107
  if (closing)
107
108
  throw new AppError(503, "Dialer is closing. No new calls are admitted.");
108
109
  const input = await body(request), key = request.headers["idempotency-key"];
110
+ if (input.transfer_enabled === true && options.allowTransfers !== true)
111
+ throw new AppError(400, "Open this dialer with transfers enabled before starting a transferable call.");
109
112
  if (typeof key !== "string" || typeof input.from_number_id !== "string" || !(await numbers()).some(number => number.id === input.from_number_id))
110
113
  throw new AppError(400, "Choose an owned voice number and a calling retry key.");
111
114
  const session = await broker.create(input, key);
112
115
  remember(session.call);
113
116
  return json(response, 200, session);
114
117
  }
118
+ const transferPath = /^\/api\/calls\/([A-Za-z0-9_-]{1,200})\/transfers(?:\/([A-Za-z0-9_-]{1,200})\/(complete|cancel))?$/.exec(url.pathname);
119
+ if (transferPath && method === "POST") {
120
+ if (closing || options.allowTransfers !== true)
121
+ throw new AppError(503, "Transfers are not enabled in this dialer.");
122
+ const id = transferPath[1], transferId = transferPath[2], operation = transferPath[3];
123
+ await own(id);
124
+ const input = await body(request);
125
+ if (transferId && (operation === "complete" || operation === "cancel"))
126
+ return json(response, 202, remember(await broker.transfers[operation](id, transferId)));
127
+ const key = request.headers["idempotency-key"];
128
+ if (typeof key !== "string" || typeof input.to !== "string" || !/^\+[1-9]\d{6,14}$/.test(input.to) || !["warm", "cold"].includes(String(input.mode)))
129
+ throw new AppError(400, "Choose a transfer destination, mode and retry key.");
130
+ return json(response, 202, remember(await broker.transfers.create(id, { to: input.to, mode: input.mode }, key)));
131
+ }
115
132
  const callPath = /^\/api\/calls\/([A-Za-z0-9_-]{1,200})(?:\/(hangup|recording))?$/.exec(url.pathname);
116
133
  if (callPath) {
117
134
  const id = callPath[1], operation = callPath[2];
@@ -15,13 +15,17 @@ export class PyaiDialerElement extends HTMLElement {
15
15
  header{display:flex;align-items:center;justify-content:space-between;gap:12px}h2{margin:0;font-size:16px;font-weight:650}.badge{font-size:11px;color:#667085;background:#f3f4f8;padding:4px 9px;border-radius:20px}
16
16
  .status{display:flex;align-items:center;gap:7px;color:#667085;margin:20px 0 5px}.dot{width:7px;height:7px;border-radius:50%;background:#9a9dad}.dot.active{background:#12a777}
17
17
  input{width:100%;border:0;border-bottom:1px solid #ececf2;padding:12px 0;font:500 26px/1.3 system-ui;letter-spacing:.5px;color:inherit;background:transparent}input:focus{outline:0;border-color:var(--dialer-accent)}
18
- .hint{font-size:12px;color:#8c91a2;margin-top:8px}.keys{display:grid;grid-template-columns:repeat(3,1fr);gap:8px;margin:22px 0}.key{background:#f8f8fb;border:0;border-radius:14px;padding:9px;cursor:pointer;font:500 23px system-ui;color:inherit;min-height:57px}.key small{display:block;font-size:9px;letter-spacing:2px;color:#9296a5;margin-top:1px}.key:hover{background:#eeedf9}
18
+ .hint,.transfer-help{font-size:12px;color:#8c91a2;margin-top:8px}.keys{display:grid;grid-template-columns:repeat(3,1fr);gap:8px;margin:22px 0}.key{background:#f8f8fb;border:0;border-radius:14px;padding:9px;cursor:pointer;font:500 23px system-ui;color:inherit;min-height:57px}.key small{display:block;font-size:9px;letter-spacing:2px;color:#9296a5;margin-top:1px}.key:hover{background:#eeedf9}
19
19
  .actions{display:flex;gap:8px}button{font:600 13px system-ui;cursor:pointer}button:focus-visible{outline:3px solid #bdb8ff;outline-offset:2px}.primary,.secondary{border:0;border-radius:13px;padding:14px 18px;min-height:46px}.primary{flex:1;color:white;background:var(--dialer-accent)}.primary.end{background:#da4055}.secondary{background:#f1f2f7;color:#535b70}button:disabled{opacity:.45;cursor:default}.error{color:#b42338;font-size:12px;min-height:18px;margin:12px 0 0}.privacy{font-size:11px;color:#8b91a1;margin:16px 0 0;text-align:center}
20
20
  @media(prefers-color-scheme:dark){:host{color:#eaebf3}section{background:#191b26;border-color:#303341}.key{background:#252836}.key:hover,.secondary{background:#34364a}.badge{background:#282b39;color:#a7aec0}input{border-color:#36394a}.error{color:#ff9eaa}}
21
21
  </style><section aria-label="Phone"><header><h2>Phone</h2><span class="badge">PyAI</span></header>
22
22
  <div class="status" role="status" aria-live="polite"><span class="dot"></span><span data-status>Connect your workspace</span></div>
23
23
  <input type="tel" inputmode="tel" autocomplete="tel" aria-label="Phone number" placeholder="+1 415 555 0100" maxlength="32"><div class="hint">Include the country code</div>
24
24
  <div class="keys"></div><div class="actions"><button class="secondary" data-action="mute" disabled>Mute</button><button class="primary" data-action="call" disabled>Call</button><button class="secondary" data-action="recover" hidden>Recover</button></div>
25
+ <div data-transfer hidden><p data-transfer-status role="status" aria-live="polite"></p><input class="transfer-number" type="tel" aria-label="Transfer phone number" placeholder="+61 2 5550 1000" maxlength="32">
26
+ <div class="actions"><button class="secondary" data-action="warm">Consult first</button><button class="secondary" data-action="cold">Transfer now</button></div>
27
+ <p class="transfer-help">Consult first keeps your caller on hold while you talk to the target. Transfer now leaves the call before the target answers.</p>
28
+ <div class="actions"><button class="primary" data-action="complete-transfer" hidden>Complete transfer</button><button class="secondary" data-action="cancel-transfer" hidden>Return to caller</button><button class="secondary" data-action="recover-transfer" hidden>Recover transfer</button></div></div>
25
29
  <p class="error" role="alert"></p><p class="privacy">Calls and recordings stay in your workspace</p></section>`;
26
30
  const letters = ["", "ABC", "DEF", "GHI", "JKL", "MNO", "PQRS", "TUV", "WXYZ", "", "+", ""];
27
31
  "123456789*0#".split("").forEach((digit, i) => {
@@ -50,6 +54,11 @@ export class PyaiDialerElement extends HTMLElement {
50
54
  this.root.querySelector("[data-action=recover]").addEventListener("click", () => void this.run(() => this.dialer.retrySetup()));
51
55
  this.root.querySelector("input").addEventListener("keydown", event => { if (event.key === "Enter")
52
56
  void this.act(); });
57
+ for (const [action, mode] of [["warm", "warm"], ["cold", "cold"]])
58
+ this.root.querySelector(`[data-action=${action}]`).addEventListener("click", () => void this.run(() => this.dialer.transfer(this.root.querySelector(".transfer-number").value.replace(/[\s()-]/g, ""), mode)));
59
+ this.root.querySelector("[data-action=complete-transfer]").addEventListener("click", () => void this.run(() => this.dialer.completeTransfer()));
60
+ this.root.querySelector("[data-action=cancel-transfer]").addEventListener("click", () => void this.run(() => this.dialer.cancelTransfer()));
61
+ this.root.querySelector("[data-action=recover-transfer]").addEventListener("click", () => void this.run(() => this.dialer.retryTransfer()));
53
62
  }
54
63
  bind(dialer, inputFor, appearance = {}, finishWrapup) {
55
64
  if (this.dialer && this.dialer !== dialer && this.dialer.current.state !== "idle")
@@ -68,7 +77,7 @@ export class PyaiDialerElement extends HTMLElement {
68
77
  this.unsubscribe = this.dialer.subscribe(s => this.render(s)); }
69
78
  disconnectedCallback() { this.unsubscribe?.(); this.unsubscribe = null; }
70
79
  render(snapshot) {
71
- const labels = { idle: "Ready to call", preparing: "Connecting…", ringing: "Ringing…", active: "Connected", stopping: "Confirming call ended…", unconfirmed: "Call needs recovery", wrapup: "Call finished" };
80
+ const labels = { idle: "Ready to call", preparing: "Connecting…", ringing: "Ringing…", active: "Connected", stopping: "Confirming call ended…", unconfirmed: "Call needs recovery", handing_off: "Transfer in progress · caller stays connected", transferred: "Transfer complete · caller stays connected", wrapup: "Call finished" };
72
81
  this.root.querySelector("[data-status]").textContent = labels[snapshot.state];
73
82
  this.root.querySelector(".dot").classList.toggle("active", snapshot.state === "active");
74
83
  const destinationInput = this.root.querySelector("input");
@@ -81,15 +90,28 @@ export class PyaiDialerElement extends HTMLElement {
81
90
  ? group.lines?.map(line => `${line.to} · ${line.status}`).join(" / ") ?? "The first answered line connects. Other lines are canceled."
82
91
  : "Include the country code";
83
92
  const call = this.root.querySelector("[data-action=call]");
84
- call.disabled = snapshot.state === "stopping";
85
- call.classList.toggle("end", !["idle", "wrapup"].includes(snapshot.state));
86
- call.textContent = snapshot.state === "idle" ? "Call" : snapshot.state === "wrapup" ? "Done" : "End call";
93
+ call.disabled = ["stopping", "handing_off"].includes(snapshot.state);
94
+ call.classList.toggle("end", !["idle", "wrapup", "transferred"].includes(snapshot.state));
95
+ call.textContent = snapshot.state === "idle" ? "Call" : ["wrapup", "transferred"].includes(snapshot.state) ? "Done" : snapshot.state === "handing_off" ? "Transferring…" : "End call";
87
96
  const mute = this.root.querySelector("[data-action=mute]");
88
97
  mute.disabled = snapshot.state !== "active";
89
98
  mute.textContent = snapshot.muted ? "Unmute" : "Mute";
90
99
  mute.setAttribute("aria-pressed", String(snapshot.muted));
91
100
  this.root.querySelector("[data-action=recover]").hidden = !["unconfirmed", "stopping"].includes(snapshot.state);
92
101
  this.root.querySelector(".error").textContent = snapshot.error ?? "";
102
+ const transfer = snapshot.call?.transfers?.at(-1), pending = transfer && !["completed", "canceled", "failed"].includes(transfer.status);
103
+ this.root.querySelector("[data-transfer]").hidden = snapshot.call?.transfer_enabled !== true || ["idle", "preparing", "ringing", "wrapup"].includes(snapshot.state);
104
+ this.root.querySelector("[data-transfer-status]").textContent = transfer ? `${transfer.to ?? "Transfer target"} · ${transfer.status}` : "Transfer this caller";
105
+ this.root.querySelector(".transfer-number").disabled = Boolean(pending || snapshot.transferPending || snapshot.state !== "active");
106
+ for (const action of ["warm", "cold"])
107
+ this.root.querySelector(`[data-action=${action}]`).disabled = Boolean(pending || snapshot.transferPending || snapshot.state !== "active");
108
+ const complete = this.root.querySelector("[data-action=complete-transfer]");
109
+ complete.hidden = transfer?.mode !== "warm" || transfer.status !== "ready";
110
+ complete.disabled = Boolean(snapshot.transferPending);
111
+ const cancel = this.root.querySelector("[data-action=cancel-transfer]");
112
+ cancel.hidden = transfer?.mode !== "warm" || !pending;
113
+ cancel.disabled = Boolean(snapshot.transferPending || ["completing", "canceling"].includes(transfer?.status ?? ""));
114
+ this.root.querySelector("[data-action=recover-transfer]").hidden = !snapshot.transferPending;
93
115
  }
94
116
  async run(action) {
95
117
  try {
@@ -103,7 +125,7 @@ export class PyaiDialerElement extends HTMLElement {
103
125
  if (!this.dialer || !this.inputFor)
104
126
  return;
105
127
  await this.run(async () => {
106
- if (this.dialer.current.state === "wrapup")
128
+ if (["wrapup", "transferred"].includes(this.dialer.current.state))
107
129
  await this.wrapup();
108
130
  else if (this.dialer.current.state !== "idle")
109
131
  await this.dialer.stop();
@@ -27,6 +27,8 @@ export class DialingQueue {
27
27
  if (!Number.isInteger(parallelLines) || parallelLines < 1 || parallelLines > 4 || mode === "parallel" && parallelLines < 2)
28
28
  throw new Error("Parallel dialing requires two to four lines.");
29
29
  if (mode === "parallel") {
30
+ if (contacts.some(contact => contact.input.transfer_enabled === true))
31
+ throw new Error("Transfer-enabled calls use single-line dialing.");
30
32
  const fingerprint = (contact) => JSON.stringify([contact.input.from_number_id, contact.input.user_id,
31
33
  contact.input.recording, contact.input.use_case ?? "support", contact.input.max_duration_seconds ?? 3600]);
32
34
  if (new Set(contacts.map(c => c.input.to)).size !== contacts.length || contacts.some(c => fingerprint(c) !== fingerprint(contacts[0])))
@@ -85,7 +87,7 @@ export class DialingQueue {
85
87
  const first = batch[0];
86
88
  this.activeId = first.id;
87
89
  this.activeIds = batch.map(contact => contact.id);
88
- const { to: _to, destinations: _destinations, ...common } = first.input;
90
+ const { to: _to, destinations: _destinations, transfer_enabled: _transferEnabled, ...common } = first.input;
89
91
  const intent = batch.length === 1 ? first.input
90
92
  : { ...common, destinations: batch.map(contact => contact.input.to) };
91
93
  // ManagedDialer persists the group and one key before any request.
@@ -40,6 +40,25 @@ export class IndexedDbDialerJournal {
40
40
  }
41
41
  if (prior?.stopping)
42
42
  next.stopping = true;
43
+ next.retiredTransferKeys = [...new Set([...(prior?.retiredTransferKeys ?? []), ...(next.retiredTransferKeys ?? [])])];
44
+ if (next.transfer && next.retiredTransferKeys.includes(next.transfer.key)) {
45
+ fail("This transfer has already finished. Refresh the original call.");
46
+ return;
47
+ }
48
+ if (prior?.transfer && !next.retiredTransferKeys.includes(prior.transfer.key)) {
49
+ if (!next.transfer)
50
+ next.transfer = structuredClone(prior.transfer);
51
+ else if (prior.transfer.key !== next.transfer.key || prior.transfer.to !== next.transfer.to || prior.transfer.mode !== next.transfer.mode ||
52
+ prior.transfer.id && next.transfer.id && prior.transfer.id !== next.transfer.id ||
53
+ prior.transfer.command && next.transfer.command && prior.transfer.command !== next.transfer.command) {
54
+ fail("Another transfer is saved for this call. Recover the original transfer.");
55
+ return;
56
+ }
57
+ else {
58
+ next.transfer.id ??= prior.transfer.id;
59
+ next.transfer.command ??= prior.transfer.command;
60
+ }
61
+ }
43
62
  store.put(next);
44
63
  result(undefined);
45
64
  };
@@ -5,6 +5,14 @@ export interface DialerBroker {
5
5
  create(input: TelephonyBrowserSessionCreate, idempotencyKey: string): Promise<TelephonyBrowserSession>;
6
6
  get(callId: string): Promise<TelephonyCall>;
7
7
  hangup(callId: string): Promise<TelephonyCall>;
8
+ transfers?: {
9
+ create(callId: string, input: {
10
+ to: string;
11
+ mode: "warm" | "cold";
12
+ }, key: string): Promise<TelephonyCall>;
13
+ complete(callId: string, transferId: string): Promise<TelephonyCall>;
14
+ cancel(callId: string, transferId: string): Promise<TelephonyCall>;
15
+ };
8
16
  }
9
17
  export interface DialerIntent {
10
18
  scope: string;
@@ -12,6 +20,14 @@ export interface DialerIntent {
12
20
  input: TelephonyBrowserSessionCreate;
13
21
  callId?: string;
14
22
  stopping: boolean;
23
+ transfer?: {
24
+ key: string;
25
+ to: string;
26
+ mode: "warm" | "cold";
27
+ id?: string;
28
+ command?: "complete" | "cancel";
29
+ };
30
+ retiredTransferKeys?: string[];
15
31
  }
16
32
  export interface DialerJournal {
17
33
  read(scope: string): Promise<DialerIntent | null>;
@@ -25,12 +41,15 @@ export interface DialerAudio {
25
41
  sendDigits(digits: string): void;
26
42
  }
27
43
  export interface DialerSnapshot {
28
- state: "idle" | "preparing" | "ringing" | "active" | "stopping" | "unconfirmed" | "wrapup";
44
+ state: "idle" | "preparing" | "ringing" | "active" | "stopping" | "unconfirmed" | "handing_off" | "transferred" | "wrapup";
29
45
  callId: string | null;
30
46
  call: TelephonyCall | null;
31
47
  muted: boolean;
32
48
  error: string | null;
49
+ transferPending?: boolean;
33
50
  }
51
+ /** A durable handoff keeps its original room after the browser goes away. */
52
+ export declare const dialerCallHandedOff: (call: TelephonyCall) => boolean;
34
53
  /** One immutable workspace/user scope per instance. Setup and stop retain the
35
54
  * original intent through lost acknowledgments, reloads and audio failures. */
36
55
  export declare class ManagedDialer {
@@ -60,6 +79,12 @@ export declare class ManagedDialer {
60
79
  stop(): Promise<void>;
61
80
  private stopKnown;
62
81
  finishWrapup(): Promise<void>;
82
+ transfer(to: string, mode: "warm" | "cold"): Promise<void>;
83
+ retryTransfer(): Promise<void>;
84
+ completeTransfer(): Promise<void>;
85
+ cancelTransfer(): Promise<void>;
86
+ private transferCommand;
87
+ private transferKnown;
63
88
  mute(value: boolean): void;
64
89
  sendDigits(digits: string): void;
65
90
  }