@mastra/livekit 0.3.0 → 0.3.1-alpha.0

package/CHANGELOG.md

@@ -1,426 +0,0 @@
-# @mastra/livekit
-
-## 0.3.0
-
-### Minor Changes
-
-- Added per-call speech-to-text and text-to-speech selection to `createLiveKitWorker`. Set the new `configuration.stt` and `configuration.tts` resolvers to pick the transcriber and voice for each call — one voice or language per tenant — keyed off the dispatch metadata and request context. Each resolver runs once per call and falls back to the top-level `stt` / `tts` option when it returns `undefined`. ([#19136](https://github.com/mastra-ai/mastra/pull/19136))
-
-  ```ts
-  export default createLiveKitWorker({
-    mastra,
-    agent: 'support',
-    stt: 'deepgram/nova-3',
-    tts: 'cartesia/sonic-3', // fallback voice
-    configuration: {
-      // Give each tenant its own voice, resolved per call from the dispatch metadata.
-      tts: ({ requestContext }) => tenantVoices[requestContext?.tenant as string],
-    },
-  });
-  ```
-
-  Previously the worker's speech pipeline was fixed at construction, so a multi-tenant worker could not vary voices or transcription per call. Customers who own their LiveKit session (the `MastraLLM` plugin path) already choose STT/TTS per call by construction; this brings the same flexibility to the batteries-included worker.
-
-- Added `MastraLLM`, a standard LiveKit LLM plugin, on the new `@mastra/livekit/plugin` entry point. Build your own `voice.AgentSession` and put a Mastra agent in the `llm` slot — the agent loop, tools, and memory run on a remote Mastra server reached over HTTP, so the worker process needs no Mastra app, database, or model provider keys. ([#19136](https://github.com/mastra-ai/mastra/pull/19136))
-
-  Before, the worker wrapper always owned the LiveKit session:
-
-  ```ts
-  import { createLiveKitWorker } from '@mastra/livekit/worker';
-  import { mastra } from './index';
-
-  export default createLiveKitWorker({
-    mastra,
-    agent: 'support',
-    stt: 'deepgram/nova-3',
-    tts: 'cartesia/sonic-3',
-  });
-  ```
-
-  Now you can own the session and keep Mastra as the LLM component:
-
-  ```ts
-  import { voice } from '@livekit/agents';
-  import { MastraLLM } from '@mastra/livekit/plugin';
-
-  const session = new voice.AgentSession({
-    llm: new MastraLLM({
-      remote: { baseUrl: process.env.MASTRA_URL!, agentId: 'support' },
-      memory: { thread: callId, resource: userId },
-    }),
-    stt: 'deepgram/nova-3',
-    tts: 'cartesia/sonic-3',
-    // Required with `memory`: LiveKit enables preemptive generation by default.
-    turnHandling: { preemptiveGeneration: { enabled: false } },
-  });
-  ```
-
-  `createLiveKitWorker` stays the batteries-included path; the plugin is the composable one. Tools keep running server-side on the Mastra agent, and interrupting the agent aborts the server-side generation.
-
-  **New transport and helpers**
-  - Added `createRemoteAgentReplyGenerator()`: streams replies from a remote Mastra server over HTTP with per-turn abort, LiveKit-typed errors, and a connect + first-token timeout. It also plugs into `createLiveKitWorker`'s `generate` option to run the existing worker against a remote server.
-  - Promoted `speakGreeting()`, `waitForAgentDoneSpeaking()`, and `runEndCall()` to public exports of `@mastra/livekit/worker`, so a worker that owns its session can rebuild the greeting and agent-initiated hang-up patterns in a few lines.
-
-  **Improvements to the existing worker**
-  - Interrupted turns now self-heal: when a caller interrupts a reply, nothing is persisted at that moment, and the part the caller actually heard is backfilled into the memory thread on the next turn — so saved transcripts match the call.
-  - Added an `onToolCall` hook that fires as each tool call starts mid-reply, the building block for tool-driven side effects such as analytics or hang-up.
-  - `onTurnComplete` now receives the turn's token usage as `result.usage`.
-
-- Added a `configuration` option to `createLiveKitWorker` — one grouped home for conversation and compliance controls, so these don't each become a separate top-level worker option. It ships with greeting/AI-disclosure controls, a consent model, and agent-initiated hang-up, and is where further compliance controls will land. ([#19136](https://github.com/mastra-ai/mastra/pull/19136))
-
-  **Greeting and AI disclosure**
-
-  `configuration.greeting` controls the opening line spoken at call start. Set `allowInterruptions: false` so a legally-required AI disclosure plays through and can't be talked over (EU AI Act Art. 50), `awaitPlayout: true` to hold post-greeting work until it finishes, and `repeatEvery` to re-disclose periodically on long calls (spoken at the next turn boundary, never mid-sentence).
-
-  ```ts
-  createLiveKitWorker({
-    mastra,
-    agent: 'support',
-    configuration: {
-      greeting: {
-        text: 'You are speaking with an AI assistant. This call may be recorded. How can I help?',
-        allowInterruptions: false,
-        awaitPlayout: true,
-        repeatEvery: 3 * 60_000, // re-disclose ~every 3 minutes
-      },
-    },
-  });
-  ```
-
-  **Per-tenant greeting**
-
-  `greeting.text` also accepts a resolver, called once per call with the call context, so one multi-tenant agent can open differently per tenant based on the dispatch metadata:
-
-  ```ts
-  greeting: {
-    text: ({ metadata }) => `Thanks for calling ${tenantName(metadata)}. You're speaking with an AI assistant.`,
-    allowInterruptions: false,
-  }
-  ```
-
-  **Consent**
-
-  `configuration.consentPolicy` declares which data-use consents a call needs, as a named, extensible set (starting with `summaryStorage`) rather than one global flag. Declaring the policy enforces nothing by itself: the new `createConsentTool` captures the caller's decision at runtime — add it to your agent and it hands each decision to your own store — and your code enforces the requirement at `onCallEnd` (or before any consent-gated step).
-
-  ```ts
-  import { createConsentTool } from '@mastra/livekit';
-
-  // in your agent's tools:
-  recordConsent: createConsentTool({
-    items: ['summaryStorage'],
-    onGrant: async ({ item, granted, resourceId }) => {
-      if (resourceId) await db.saveConsent(resourceId, item, granted);
-    },
-  }),
-  ```
-
-  **Agent-initiated hang-up**
-
-  `configuration.endCall` lets the agent end the call itself. Add the new `createEndCallTool` to your agent and instruct it to say goodbye and then call the tool; the worker waits for the closing words to finish playing, holds a short audio drain (`drainMs`, default 800ms) so the tail of the goodbye isn't clipped while it's still buffered at the caller, then hangs up — running `onCallEnd` on the way out, exactly as a caller hang-up does. It works on both the agent and workflow reply paths.
-
-  ```ts
-  import { createEndCallTool } from '@mastra/livekit';
-
-  // in your agent's tools:
-  endCall: (createEndCallTool(),
-    // on the worker:
-    createLiveKitWorker({ mastra, agent: 'support', configuration: { endCall: {} } }));
-  ```
-
-  **Backwards compatible**
-
-  The previous top-level `greeting` (string) and `persistGreeting` options still work as deprecated aliases for `configuration.greeting.text` and `configuration.greeting.persist`. When both are set, `configuration.greeting` wins field by field, so existing worker configs keep running unchanged.
-
-### Patch Changes
-
-- Updated dependencies [[`bd6d240`](https://github.com/mastra-ai/mastra/commit/bd6d2402db93dddaef0721667e7e8a030e7c6e16), [`0111486`](https://github.com/mastra-ai/mastra/commit/01114867612593eef5cfa2fda6a1194dfedda841), [`96a3749`](https://github.com/mastra-ai/mastra/commit/96a37492235f5b8076b3e3177d83ed5a5e44a640), [`fe1bda0`](https://github.com/mastra-ai/mastra/commit/fe1bda06f6af92a694a51712db747cda1e7185f0), [`25e7c12`](https://github.com/mastra-ai/mastra/commit/25e7c126a770069ae7fb7ecf1d2adb40e017b009), [`1ce5121`](https://github.com/mastra-ai/mastra/commit/1ce512155d122bb21f47d98383e82ffbf84b39e8), [`fb8aea3`](https://github.com/mastra-ai/mastra/commit/fb8aea384291e77311be3a64ee1717320d5c3c73), [`4adc391`](https://github.com/mastra-ai/mastra/commit/4adc3911075249c352bb4832d2471922826344de), [`a5c6337`](https://github.com/mastra-ai/mastra/commit/a5c6337d23c7686c81a32ce62f550f610543a240), [`3cfc47a`](https://github.com/mastra-ai/mastra/commit/3cfc47a6b89940aadd0f46fb01ae9624a73a865d), [`2bb7817`](https://github.com/mastra-ai/mastra/commit/2bb78176112fde628483de2830528f7eee911e56), [`51d9870`](https://github.com/mastra-ai/mastra/commit/51d987032c689c2855374d0f244f5d654da809d1), [`5cab274`](https://github.com/mastra-ai/mastra/commit/5cab2744250e22d12fefa7b32637dce224233cee), [`7fa27d3`](https://github.com/mastra-ai/mastra/commit/7fa27d3b6f5ed68cd34e454a4d3ad9c482a0cfbc), [`8b97958`](https://github.com/mastra-ai/mastra/commit/8b979589f9aa59ba67cac565949475f2ffeb4ac3), [`8410541`](https://github.com/mastra-ai/mastra/commit/84105412c60ecd3bb33a9838146f59c4b588228f), [`a58dcbb`](https://github.com/mastra-ai/mastra/commit/a58dcbb546d7e1d65ebdc1f39e55f0908fcd9391), [`aa38805`](https://github.com/mastra-ai/mastra/commit/aa38805b878b827403be785eb90688d7172f5a40), [`153bd3b`](https://github.com/mastra-ai/mastra/commit/153bd3b396bdfed6b74cf43de12db8fd2d83c04a), [`45a8e65`](https://github.com/mastra-ai/mastra/commit/45a8e65e1556d1362cb3f25187023c36de26661d), [`e955965`](https://github.com/mastra-ai/mastra/commit/e955965dce575a903e37cf054d28ea99aa48785e), [`2d22570`](https://github.com/mastra-ai/mastra/commit/2d22570c7dfdd02123d0ecc529efb05ccba2d9fc), [`07bb863`](https://github.com/mastra-ai/mastra/commit/07bb8631919c6f7cf377dccd45b096e0f17fbed0), [`c8ed116`](https://github.com/mastra-ai/mastra/commit/c8ed11699f62bcac70102ab4ec84d80d20541da6), [`01b338c`](https://github.com/mastra-ai/mastra/commit/01b338c56271f0219606710e3e8b26dee27ac6c2), [`a99eae8`](https://github.com/mastra-ai/mastra/commit/a99eae8908e500c1b2d12f9d277be616b98617a5), [`860ef7e`](https://github.com/mastra-ai/mastra/commit/860ef7e77d92b63469cbe5857aa1e626197e43e9), [`17e818c`](https://github.com/mastra-ai/mastra/commit/17e818c51a958ba90641b1a959dc38faf8c034e9), [`edce8d2`](https://github.com/mastra-ai/mastra/commit/edce8d2769f19e27a05737c627af2d765472a4f8), [`8a586ec`](https://github.com/mastra-ai/mastra/commit/8a586eca9a4914f31dff6140d0d45ac375b00669), [`4451dfe`](https://github.com/mastra-ai/mastra/commit/4451dfe857428e7abcc0261a507a2e186dae6d47), [`8b7361d`](https://github.com/mastra-ai/mastra/commit/8b7361d35de68b80d05d30a74e0c69e7218fd612), [`1d39058`](https://github.com/mastra-ai/mastra/commit/1d39058e548efd691799985d5c8af2737f1c3bd2), [`3927473`](https://github.com/mastra-ai/mastra/commit/392747323ddb10c643d12be7b9ae913159dfaeed), [`dce50dc`](https://github.com/mastra-ai/mastra/commit/dce50dc9a1c1fcd0f427bb5f6250ec74910cb04b), [`fd13f8e`](https://github.com/mastra-ai/mastra/commit/fd13f8e21990f9904c3eedba3a626bb4a929cdb8), [`634caff`](https://github.com/mastra-ai/mastra/commit/634caff29a9200ad058b67d53f96d9e5832fb8a2), [`f703f87`](https://github.com/mastra-ai/mastra/commit/f703f878de072d51fda557f9c50867d8252bef05), [`3e26c87`](https://github.com/mastra-ai/mastra/commit/3e26c87de0c5bc2583b795ce6ca5889b6b161acb), [`33f2b88`](https://github.com/mastra-ai/mastra/commit/33f2b88842c09a567f906fac4cb61cd5277ced59), [`177010f`](https://github.com/mastra-ai/mastra/commit/177010ff096d2e4b28d89803be5b1a4cad2a0d6b), [`0ad646f`](https://github.com/mastra-ai/mastra/commit/0ad646f71a530f2454664299e5e01bfd13fa12e5), [`b486abf`](https://github.com/mastra-ai/mastra/commit/b486abfa2a7528c6f527e4015c819ea9fa54aaad), [`54a51e0`](https://github.com/mastra-ai/mastra/commit/54a51e0a484fe1ebad3fb1f7ef5282a075709eb7), [`c43f3a9`](https://github.com/mastra-ai/mastra/commit/c43f3a9d1efde99b38789364ba4d0ba670f430e3), [`a5008f2`](https://github.com/mastra-ai/mastra/commit/a5008f22ae710ad9402ea9f2547d8c02f74d384b), [`e2d5f37`](https://github.com/mastra-ai/mastra/commit/e2d5f373bd289be534d5f8694d34465010533df6), [`4ce0163`](https://github.com/mastra-ai/mastra/commit/4ce0163dc86e675a86809685c8ce6c49f1aeb87e), [`4378341`](https://github.com/mastra-ai/mastra/commit/43783412df5ea3dd35f5b1f6e4851e79c346fc89)]:
-  - @mastra/core@1.51.0
-
-## 0.3.0-alpha.0
-
-### Minor Changes
-
-- Added per-call speech-to-text and text-to-speech selection to `createLiveKitWorker`. Set the new `configuration.stt` and `configuration.tts` resolvers to pick the transcriber and voice for each call — one voice or language per tenant — keyed off the dispatch metadata and request context. Each resolver runs once per call and falls back to the top-level `stt` / `tts` option when it returns `undefined`. ([#19136](https://github.com/mastra-ai/mastra/pull/19136))
-
-  ```ts
-  export default createLiveKitWorker({
-    mastra,
-    agent: 'support',
-    stt: 'deepgram/nova-3',
-    tts: 'cartesia/sonic-3', // fallback voice
-    configuration: {
-      // Give each tenant its own voice, resolved per call from the dispatch metadata.
-      tts: ({ requestContext }) => tenantVoices[requestContext?.tenant as string],
-    },
-  });
-  ```
-
-  Previously the worker's speech pipeline was fixed at construction, so a multi-tenant worker could not vary voices or transcription per call. Customers who own their LiveKit session (the `MastraLLM` plugin path) already choose STT/TTS per call by construction; this brings the same flexibility to the batteries-included worker.
-
-- Added `MastraLLM`, a standard LiveKit LLM plugin, on the new `@mastra/livekit/plugin` entry point. Build your own `voice.AgentSession` and put a Mastra agent in the `llm` slot — the agent loop, tools, and memory run on a remote Mastra server reached over HTTP, so the worker process needs no Mastra app, database, or model provider keys. ([#19136](https://github.com/mastra-ai/mastra/pull/19136))
-
-  Before, the worker wrapper always owned the LiveKit session:
-
-  ```ts
-  import { createLiveKitWorker } from '@mastra/livekit/worker';
-  import { mastra } from './index';
-
-  export default createLiveKitWorker({
-    mastra,
-    agent: 'support',
-    stt: 'deepgram/nova-3',
-    tts: 'cartesia/sonic-3',
-  });
-  ```
-
-  Now you can own the session and keep Mastra as the LLM component:
-
-  ```ts
-  import { voice } from '@livekit/agents';
-  import { MastraLLM } from '@mastra/livekit/plugin';
-
-  const session = new voice.AgentSession({
-    llm: new MastraLLM({
-      remote: { baseUrl: process.env.MASTRA_URL!, agentId: 'support' },
-      memory: { thread: callId, resource: userId },
-    }),
-    stt: 'deepgram/nova-3',
-    tts: 'cartesia/sonic-3',
-    // Required with `memory`: LiveKit enables preemptive generation by default.
-    turnHandling: { preemptiveGeneration: { enabled: false } },
-  });
-  ```
-
-  `createLiveKitWorker` stays the batteries-included path; the plugin is the composable one. Tools keep running server-side on the Mastra agent, and interrupting the agent aborts the server-side generation.
-
-  **New transport and helpers**
-  - Added `createRemoteAgentReplyGenerator()`: streams replies from a remote Mastra server over HTTP with per-turn abort, LiveKit-typed errors, and a connect + first-token timeout. It also plugs into `createLiveKitWorker`'s `generate` option to run the existing worker against a remote server.
-  - Promoted `speakGreeting()`, `waitForAgentDoneSpeaking()`, and `runEndCall()` to public exports of `@mastra/livekit/worker`, so a worker that owns its session can rebuild the greeting and agent-initiated hang-up patterns in a few lines.
-
-  **Improvements to the existing worker**
-  - Interrupted turns now self-heal: when a caller interrupts a reply, nothing is persisted at that moment, and the part the caller actually heard is backfilled into the memory thread on the next turn — so saved transcripts match the call.
-  - Added an `onToolCall` hook that fires as each tool call starts mid-reply, the building block for tool-driven side effects such as analytics or hang-up.
-  - `onTurnComplete` now receives the turn's token usage as `result.usage`.
-
-- Added a `configuration` option to `createLiveKitWorker` — one grouped home for conversation and compliance controls, so these don't each become a separate top-level worker option. It ships with greeting/AI-disclosure controls, a consent model, and agent-initiated hang-up, and is where further compliance controls will land. ([#19136](https://github.com/mastra-ai/mastra/pull/19136))
-
-  **Greeting and AI disclosure**
-
-  `configuration.greeting` controls the opening line spoken at call start. Set `allowInterruptions: false` so a legally-required AI disclosure plays through and can't be talked over (EU AI Act Art. 50), `awaitPlayout: true` to hold post-greeting work until it finishes, and `repeatEvery` to re-disclose periodically on long calls (spoken at the next turn boundary, never mid-sentence).
-
-  ```ts
-  createLiveKitWorker({
-    mastra,
-    agent: 'support',
-    configuration: {
-      greeting: {
-        text: 'You are speaking with an AI assistant. This call may be recorded. How can I help?',
-        allowInterruptions: false,
-        awaitPlayout: true,
-        repeatEvery: 3 * 60_000, // re-disclose ~every 3 minutes
-      },
-    },
-  });
-  ```
-
-  **Per-tenant greeting**
-
-  `greeting.text` also accepts a resolver, called once per call with the call context, so one multi-tenant agent can open differently per tenant based on the dispatch metadata:
-
-  ```ts
-  greeting: {
-    text: ({ metadata }) => `Thanks for calling ${tenantName(metadata)}. You're speaking with an AI assistant.`,
-    allowInterruptions: false,
-  }
-  ```
-
-  **Consent**
-
-  `configuration.consentPolicy` declares which data-use consents a call needs, as a named, extensible set (starting with `summaryStorage`) rather than one global flag. Declaring the policy enforces nothing by itself: the new `createConsentTool` captures the caller's decision at runtime — add it to your agent and it hands each decision to your own store — and your code enforces the requirement at `onCallEnd` (or before any consent-gated step).
-
-  ```ts
-  import { createConsentTool } from '@mastra/livekit';
-
-  // in your agent's tools:
-  recordConsent: createConsentTool({
-    items: ['summaryStorage'],
-    onGrant: async ({ item, granted, resourceId }) => {
-      if (resourceId) await db.saveConsent(resourceId, item, granted);
-    },
-  }),
-  ```
-
-  **Agent-initiated hang-up**
-
-  `configuration.endCall` lets the agent end the call itself. Add the new `createEndCallTool` to your agent and instruct it to say goodbye and then call the tool; the worker waits for the closing words to finish playing, holds a short audio drain (`drainMs`, default 800ms) so the tail of the goodbye isn't clipped while it's still buffered at the caller, then hangs up — running `onCallEnd` on the way out, exactly as a caller hang-up does. It works on both the agent and workflow reply paths.
-
-  ```ts
-  import { createEndCallTool } from '@mastra/livekit';
-
-  // in your agent's tools:
-  endCall: (createEndCallTool(),
-    // on the worker:
-    createLiveKitWorker({ mastra, agent: 'support', configuration: { endCall: {} } }));
-  ```
-
-  **Backwards compatible**
-
-  The previous top-level `greeting` (string) and `persistGreeting` options still work as deprecated aliases for `configuration.greeting.text` and `configuration.greeting.persist`. When both are set, `configuration.greeting` wins field by field, so existing worker configs keep running unchanged.
-
-## 0.2.0
-
-### Minor Changes
-
-- Added `@mastra/livekit`, a new package that turns Mastra agents into realtime voice agents using LiveKit. ([#17896](https://github.com/mastra-ai/mastra/pull/17896))
-
-  LiveKit's agents framework runs the audio loop — WebRTC transport, voice activity detection, streaming speech-to-text, semantic turn detection, and barge-in — while your Mastra agent generates every reply with its own model, tools, and memory. When a caller interrupts the agent, LiveKit cancels the in-flight stream and Mastra stops generating.
-
-  **Build a voice worker**
-  - `createLiveKitWorker()` builds a LiveKit worker that answers voice sessions with your Mastra agents; `runLiveKitWorker()` starts its CLI (`dev`/`start`). Both live on the `@mastra/livekit/worker` entry point.
-  - `liveKitConnectionRoute()` is an API route that mints LiveKit tokens and dispatches the voice agent into a room; `dispatchVoiceSession()` does the same programmatically for server-initiated sessions like outbound calls. These live on the `@mastra/livekit` entry point, which is safe to import from Mastra server code — it never loads the LiveKit agents runtime.
-
-  ```ts
-  // src/mastra/voice-worker.ts
-  import { createLiveKitWorker } from '@mastra/livekit/worker';
-  import { mastra } from './index';
-
-  export default createLiveKitWorker({
-    mastra,
-    agent: 'support',
-    stt: 'deepgram/nova-3',
-    tts: 'cartesia/sonic-3',
-    turnDetection: 'multilingual',
-  });
-  ```
-
-  **Drive replies with an agent or a workflow**
-
-  Each turn's reply can come from a Mastra agent (the default) or a Mastra workflow. With a workflow, LiveKit still owns the audio loop and calls into Mastra once per turn, so the workflow runs to completion each turn (no suspend/resume) — pass the transcript in, stream the reply out.
-  - `workflow` / `workflowInput` options on `createLiveKitWorker()` drive replies with a workflow.
-  - `pipeAgentReplyToWriter(agentStream, writer)` streams an agent's reply from inside a workflow step, forwarding both its words and its tool calls (piping only the text would drop the tool calls).
-  - `generate` is an escape hatch to plug in any custom reply generator.
-
-  ```ts
-  export default createLiveKitWorker({
-    mastra,
-    workflow: 'phoneConversation',
-    workflowInput: ({ messages }) => ({ turn: messages }),
-    replyStep: 'generateResponse',
-    stt: 'deepgram/nova-3',
-    tts: 'cartesia/sonic-3',
-  });
-  ```
-
-  **Run work after each turn and at the end of the call**
-  - `onTurnComplete` runs once per turn, right after the reply finishes playing. It runs in the background — the worker never waits for it — so you can save memory, update your CRM, or record analytics without adding any delay for the caller or the next reply. It also runs with `result.interrupted: true` when the caller talks over the agent.
-  - `onCallEnd` runs once when the call ends. Unlike `onTurnComplete`, the worker waits for it to finish before exiting, so it's the place for end-of-call work like summarizing the whole conversation into long-term memory once.
-  - `toolFeedback` speaks a short phrase while a tool runs; `memoryInstance` gives the workflow path a `Memory` instance to open the call's thread and save the greeting, so the saved conversation is complete — greeting included — like the agent path.
-
-  Both hooks work whether you drive replies with an agent or a workflow.
-
-  ```ts
-  createLiveKitWorker({
-    mastra,
-    agent: 'callCenter',
-    onTurnComplete: async ({ result, memory }) => {
-      if (memory) await crm.logContact(memory.resource, result.text);
-    },
-    onCallEnd: async ({ memory }) => {
-      // After the caller hangs up: save a lasting summary of the call.
-    },
-  });
-  ```
-
-  **Built-in observability**
-
-  When the Mastra instance has observability configured, each call opens a `voice call` trace that nests every turn's agent run and adds child spans for LiveKit's speech-to-text, text-to-speech, turn-detection, and LLM latency, closing with a per-model token, character, and audio usage roll-up. On by default; pass `observability: false` to disable.
-
-  **Studio voice mode**
-
-  Studio's agent chat gains a voice call mode: when the Mastra server exposes a LiveKit connection route and a voice worker is running, a phone button in the chat composer starts a realtime voice session with the agent. Live captions, agent state (listening, thinking, speaking), and barge-in all surface in the chat, and the conversation lands in the same memory thread as text chat.
-
-  See the [LiveKit voice guide](https://mastra.ai/docs/voice/livekit) for setup.
-
-### Patch Changes
-
-- Updated dependencies [[`b291760`](https://github.com/mastra-ai/mastra/commit/b291760df9d6c7e4fc72606c8f0a4af2cf6e946c), [`3ffb8b7`](https://github.com/mastra-ai/mastra/commit/3ffb8b720e90f5e6977129ec1f6707d43c2bebe0), [`6ef59fe`](https://github.com/mastra-ai/mastra/commit/6ef59fef1da52ed8da5fbb2a892c71cf4fb6c739), [`4039488`](https://github.com/mastra-ai/mastra/commit/403948898af7293198d9e8b3e7fb47f623c78b94), [`29b7ea6`](https://github.com/mastra-ai/mastra/commit/29b7ea64e72b5523d5bdcbd34ee03d2b854d54e1), [`b2c9d70`](https://github.com/mastra-ai/mastra/commit/b2c9d70757207fb01a9069549e69b6f0d73a6636), [`a51c63d`](https://github.com/mastra-ai/mastra/commit/a51c63d8ee639e4daeba2a0be093efa6a1b5e52f), [`252f63d`](https://github.com/mastra-ai/mastra/commit/252f63d8fec723955adb2202be2f01a75ad0e69c), [`5ea76a7`](https://github.com/mastra-ai/mastra/commit/5ea76a723d966c72da9aa3ab30ae20276e049765), [`6445560`](https://github.com/mastra-ai/mastra/commit/6445560327045d20b239585fc63fed72e9ce36ec), [`e2b9f33`](https://github.com/mastra-ai/mastra/commit/e2b9f33456fd638eca555f9466c6519d8d049666), [`10959d5`](https://github.com/mastra-ai/mastra/commit/10959d509d824f682d40ff96e05ee044aec3b0e5), [`c547a77`](https://github.com/mastra-ai/mastra/commit/c547a7729bdf64dfc2df29c965046c0712a18f10), [`a0085fa`](https://github.com/mastra-ai/mastra/commit/a0085fa0934e52c37c8c8b3d75a6bb5cd199af36), [`a2ba369`](https://github.com/mastra-ai/mastra/commit/a2ba369e796dfab610f41c6875965b488272fa55), [`ffc3c17`](https://github.com/mastra-ai/mastra/commit/ffc3c17274ea17c11aa6f73d3140649cd7fc8abc), [`81542c1`](https://github.com/mastra-ai/mastra/commit/81542c1835c35bc32f2ce4fa9136ee11993cd299), [`3908e53`](https://github.com/mastra-ai/mastra/commit/3908e53ce04bbea04f5e0c097d7aa298c35fabee), [`cb24ce7`](https://github.com/mastra-ai/mastra/commit/cb24ce76bd16ca88eb6a963f6277f8780e703029), [`02705fd`](https://github.com/mastra-ai/mastra/commit/02705fd2f5a9062210d64ea061adeeb10dc9452e), [`ae51e81`](https://github.com/mastra-ai/mastra/commit/ae51e818825582d42500338dfc1929a082eff0ba), [`6f304ef`](https://github.com/mastra-ai/mastra/commit/6f304ef319e99725e884bdb8d3193c001b6e5964), [`5f9858f`](https://github.com/mastra-ai/mastra/commit/5f9858f791f1137ca7d52d23559fb4568f7a9026)]:
-  - @mastra/core@1.50.0
-
-## 0.2.0-alpha.0
-
-### Minor Changes
-
-- Added `@mastra/livekit`, a new package that turns Mastra agents into realtime voice agents using LiveKit. ([#17896](https://github.com/mastra-ai/mastra/pull/17896))
-
-  LiveKit's agents framework runs the audio loop — WebRTC transport, voice activity detection, streaming speech-to-text, semantic turn detection, and barge-in — while your Mastra agent generates every reply with its own model, tools, and memory. When a caller interrupts the agent, LiveKit cancels the in-flight stream and Mastra stops generating.
-
-  **Build a voice worker**
-  - `createLiveKitWorker()` builds a LiveKit worker that answers voice sessions with your Mastra agents; `runLiveKitWorker()` starts its CLI (`dev`/`start`). Both live on the `@mastra/livekit/worker` entry point.
-  - `liveKitConnectionRoute()` is an API route that mints LiveKit tokens and dispatches the voice agent into a room; `dispatchVoiceSession()` does the same programmatically for server-initiated sessions like outbound calls. These live on the `@mastra/livekit` entry point, which is safe to import from Mastra server code — it never loads the LiveKit agents runtime.
-
-  ```ts
-  // src/mastra/voice-worker.ts
-  import { createLiveKitWorker } from '@mastra/livekit/worker';
-  import { mastra } from './index';
-
-  export default createLiveKitWorker({
-    mastra,
-    agent: 'support',
-    stt: 'deepgram/nova-3',
-    tts: 'cartesia/sonic-3',
-    turnDetection: 'multilingual',
-  });
-  ```
-
-  **Drive replies with an agent or a workflow**
-
-  Each turn's reply can come from a Mastra agent (the default) or a Mastra workflow. With a workflow, LiveKit still owns the audio loop and calls into Mastra once per turn, so the workflow runs to completion each turn (no suspend/resume) — pass the transcript in, stream the reply out.
-  - `workflow` / `workflowInput` options on `createLiveKitWorker()` drive replies with a workflow.
-  - `pipeAgentReplyToWriter(agentStream, writer)` streams an agent's reply from inside a workflow step, forwarding both its words and its tool calls (piping only the text would drop the tool calls).
-  - `generate` is an escape hatch to plug in any custom reply generator.
-
-  ```ts
-  export default createLiveKitWorker({
-    mastra,
-    workflow: 'phoneConversation',
-    workflowInput: ({ messages }) => ({ turn: messages }),
-    replyStep: 'generateResponse',
-    stt: 'deepgram/nova-3',
-    tts: 'cartesia/sonic-3',
-  });
-  ```
-
-  **Run work after each turn and at the end of the call**
-  - `onTurnComplete` runs once per turn, right after the reply finishes playing. It runs in the background — the worker never waits for it — so you can save memory, update your CRM, or record analytics without adding any delay for the caller or the next reply. It also runs with `result.interrupted: true` when the caller talks over the agent.
-  - `onCallEnd` runs once when the call ends. Unlike `onTurnComplete`, the worker waits for it to finish before exiting, so it's the place for end-of-call work like summarizing the whole conversation into long-term memory once.
-  - `toolFeedback` speaks a short phrase while a tool runs; `memoryInstance` gives the workflow path a `Memory` instance to open the call's thread and save the greeting, so the saved conversation is complete — greeting included — like the agent path.
-
-  Both hooks work whether you drive replies with an agent or a workflow.
-
-  ```ts
-  createLiveKitWorker({
-    mastra,
-    agent: 'callCenter',
-    onTurnComplete: async ({ result, memory }) => {
-      if (memory) await crm.logContact(memory.resource, result.text);
-    },
-    onCallEnd: async ({ memory }) => {
-      // After the caller hangs up: save a lasting summary of the call.
-    },
-  });
-  ```
-
-  **Built-in observability**
-
-  When the Mastra instance has observability configured, each call opens a `voice call` trace that nests every turn's agent run and adds child spans for LiveKit's speech-to-text, text-to-speech, turn-detection, and LLM latency, closing with a per-model token, character, and audio usage roll-up. On by default; pass `observability: false` to disable.
-
-  **Studio voice mode**
-
-  Studio's agent chat gains a voice call mode: when the Mastra server exposes a LiveKit connection route and a voice worker is running, a phone button in the chat composer starts a realtime voice session with the agent. Live captions, agent state (listening, thinking, speaking), and barge-in all surface in the chat, and the conversation lands in the same memory thread as text chat.
-
-  See the [LiveKit voice guide](https://mastra.ai/docs/voice/livekit) for setup.
-
-### Patch Changes
-
-- Updated dependencies [[`a0085fa`](https://github.com/mastra-ai/mastra/commit/a0085fa0934e52c37c8c8b3d75a6bb5cd199af36)]:
-  - @mastra/core@1.50.0-alpha.5

package/dist/chunk-2E3MTAOA.js

@@ -1,133 +0,0 @@
-import { ReadableStream } from 'stream/web';
-
-// src/constants.ts
-var DEFAULT_LIVEKIT_AGENT_NAME = "mastra-voice";
-
-// src/metadata.ts
-function parseSessionMetadata(raw) {
-  if (!raw) return {};
-  try {
-    const parsed = JSON.parse(raw);
-    if (parsed && typeof parsed === "object" && !Array.isArray(parsed)) {
-      return parsed;
-    }
-  } catch {
-  }
-  return {};
-}
-function serializeSessionMetadata(metadata) {
-  return JSON.stringify(metadata);
-}
-function unwrapStepText(output) {
-  if (typeof output === "string") return output;
-  if (output && typeof output === "object" && output.type === "text-delta") {
-    const text = output.payload?.text;
-    return typeof text === "string" ? text : void 0;
-  }
-  return void 0;
-}
-function unwrapStepToolCall(output) {
-  if (!output || typeof output !== "object" || output.type !== "tool-call") return void 0;
-  const payload = output.payload;
-  if (payload && typeof payload.toolCallId === "string" && typeof payload.toolName === "string") {
-    return { toolCallId: payload.toolCallId, toolName: payload.toolName, args: payload.args };
-  }
-  return void 0;
-}
-function pipeAgentReplyToWriter(agentStream, writer) {
-  let text = "";
-  const forwarded = new ReadableStream({
-    start: async (controller) => {
-      for await (const chunk of agentStream.fullStream) {
-        const type = chunk?.type;
-        if (type === "text-delta") {
-          const delta = chunk.payload?.text;
-          if (typeof delta === "string" && delta) {
-            text += delta;
-            controller.enqueue(chunk);
-          }
-        } else if (type === "tool-call") {
-          controller.enqueue(chunk);
-        }
-      }
-      controller.close();
-    }
-  });
-  return forwarded.pipeTo(writer).then(() => text);
-}
-function createWorkflowReplyGenerator(options) {
-  const { workflow, workflowInput, replyStep, resultText, toolFeedback, onTurnComplete } = options;
-  return async (ctx) => {
-    const inputData = await workflowInput(ctx);
-    const run = await workflow.createRun();
-    const streamArgs = {
-      inputData
-    };
-    if (ctx.tracingContext) streamArgs.tracingContext = ctx.tracingContext;
-    if (ctx.requestContext) streamArgs.requestContext = ctx.requestContext;
-    const output = run.stream(streamArgs);
-    let cancelled = false;
-    let replyText = "";
-    const toolCalls = [];
-    const emitTurnComplete = (interrupted) => {
-      if (!onTurnComplete) return;
-      const completeCtx = { ...ctx, result: { text: replyText, toolCalls, interrupted } };
-      Promise.resolve().then(() => onTurnComplete(completeCtx)).catch((error) => {
-        console.warn("@mastra/livekit: onTurnComplete hook threw", error);
-      });
-    };
-    return new ReadableStream({
-      start: async (controller) => {
-        let streamedAny = false;
-        try {
-          for await (const chunk of output.fullStream) {
-            if (cancelled) break;
-            if (chunk.type !== "workflow-step-output") continue;
-            const payload = chunk.payload;
-            if (replyStep && payload.stepName !== replyStep) continue;
-            const text = unwrapStepText(payload.output);
-            if (text) {
-              streamedAny = true;
-              replyText += text;
-              controller.enqueue(text);
-              continue;
-            }
-            const toolCall = unwrapStepToolCall(payload.output);
-            if (toolCall) {
-              toolCalls.push(toolCall);
-              if (toolFeedback) {
-                const filler = toolFeedback(toolCall);
-                if (filler) controller.enqueue(filler.endsWith(" ") ? filler : `${filler} `);
-              }
-            }
-          }
-          if (!cancelled && !streamedAny && resultText) {
-            const finalText = resultText(await output.result);
-            if (finalText) {
-              replyText += finalText;
-              controller.enqueue(finalText);
-            }
-          }
-          if (!cancelled) controller.close();
-          emitTurnComplete(cancelled);
-        } catch (error) {
-          if (cancelled) {
-            emitTurnComplete(true);
-            return;
-          }
-          controller.error(error);
-        }
-      },
-      cancel: () => {
-        cancelled = true;
-        void Promise.resolve(run.cancel()).catch((error) => {
-          console.warn("@mastra/livekit: failed to cancel the workflow run on barge-in", error);
-        });
-      }
-    });
-  };
-}
-
-export { DEFAULT_LIVEKIT_AGENT_NAME, createWorkflowReplyGenerator, parseSessionMetadata, pipeAgentReplyToWriter, serializeSessionMetadata };
-//# sourceMappingURL=chunk-2E3MTAOA.js.map
-//# sourceMappingURL=chunk-2E3MTAOA.js.map

package/dist/chunk-2E3MTAOA.js.map

@@ -1 +0,0 @@
-{"version":3,"sources":["../src/constants.ts","../src/metadata.ts","../src/workflow-generator.ts"],"names":[],"mappings":";;;AACO,IAAM,0BAAA,GAA6B;;;ACcnC,SAAS,qBAAqB,GAAA,EAAwD;AAC3F,EAAA,IAAI,CAAC,GAAA,EAAK,OAAO,EAAC;AAClB,EAAA,IAAI;AACF,IAAA,MAAM,MAAA,GAAS,IAAA,CAAK,KAAA,CAAM,GAAG,CAAA;AAC7B,IAAA,IAAI,MAAA,IAAU,OAAO,MAAA,KAAW,QAAA,IAAY,CAAC,KAAA,CAAM,OAAA,CAAQ,MAAM,CAAA,EAAG;AAClE,MAAA,OAAO,MAAA;AAAA,IACT;AAAA,EACF,CAAA,CAAA,MAAQ;AAAA,EAER;AACA,EAAA,OAAO,EAAC;AACV;AAEO,SAAS,yBAAyB,QAAA,EAA0C;AACjF,EAAA,OAAO,IAAA,CAAK,UAAU,QAAQ,CAAA;AAChC;ACsBO,SAAS,eAAe,MAAA,EAAqC;AAClE,EAAA,IAAI,OAAO,MAAA,KAAW,QAAA,EAAU,OAAO,MAAA;AACvC,EAAA,IAAI,UAAU,OAAO,MAAA,KAAW,QAAA,IAAa,MAAA,CAA8B,SAAS,YAAA,EAAc;AAChG,IAAA,MAAM,IAAA,GAAQ,OAA4C,OAAA,EAAS,IAAA;AACnE,IAAA,OAAO,OAAO,IAAA,KAAS,QAAA,GAAW,IAAA,GAAO,MAAA;AAAA,EAC3C;AACA,EAAA,OAAO,MAAA;AACT;AAOO,SAAS,mBAAmB,MAAA,EAA4C;AAC7E,EAAA,IAAI,CAAC,UAAU,OAAO,MAAA,KAAW,YAAa,MAAA,CAA8B,IAAA,KAAS,aAAa,OAAO,MAAA;AACzG,EAAA,MAAM,UAAW,MAAA,CAAsF,OAAA;AACvG,EAAA,IAAI,OAAA,IAAW,OAAO,OAAA,CAAQ,UAAA,KAAe,YAAY,OAAO,OAAA,CAAQ,aAAa,QAAA,EAAU;AAC7F,IAAA,OAAO,EAAE,YAAY,OAAA,CAAQ,UAAA,EAAY,UAAU,OAAA,CAAQ,QAAA,EAAU,IAAA,EAAM,OAAA,CAAQ,IAAA,EAAK;AAAA,EAC1F;AACA,EAAA,OAAO,MAAA;AACT;AA+BO,SAAS,sBAAA,CACd,aACA,MAAA,EACiB;AACjB,EAAA,IAAI,IAAA,GAAO,EAAA;AAGX,EAAA,MAAM,SAAA,GAAY,IAAI,cAAA,CAAwB;AAAA,IAC5C,KAAA,EAAO,OAAM,UAAA,KAAc;AACzB,MAAA,WAAA,MAAiB,KAAA,IAAS,YAAY,UAAA,EAAY;AAChD,QAAA,MAAM,OAAQ,KAAA,EAA8B,IAAA;AAC5C,QAAA,IAAI,SAAS,YAAA,EAAc;AACzB,UAAA,MAAM,KAAA,GAAS,MAA2C,OAAA,EAAS,IAAA;AACnE,UAAA,IAAI,OAAO,KAAA,KAAU,QAAA,IAAY,KAAA,EAAO;AACtC,YAAA,IAAA,IAAQ,KAAA;AACR,YAAA,UAAA,CAAW,QAAQ,KAAK,CAAA;AAAA,UAC1B;AAAA,QACF,CAAA,MAAA,IAAW,SAAS,WAAA,EAAa;AAC/B,UAAA,UAAA,CAAW,QAAQ,KAAK,CAAA;AAAA,QAC1B;AAAA,MACF;AACA,MAAA,UAAA,CAAW,KAAA,EAAM;AAAA,IACnB;AAAA,GACD,CAAA;AACD,EAAA,OAAO,UAAU,MAAA,CAAO,MAAM,CAAA,CAAE,IAAA,CAAK,MAAM,IAAI,CAAA;AACjD;AAeO,SAAS,6BAA6B,OAAA,EAA6D;AACxG,EAAA,MAAM,EAAE,QAAA,EAAU,aAAA,EAAe,WAAW,UAAA,EAAY,YAAA,EAAc,gBAAe,GAAI,OAAA;AACzF,EAAA,OAAO,OAAM,GAAA,KAAO;AAClB,IAAA,MAAM,SAAA,GAAY,MAAM,aAAA,CAAc,GAAG,CAAA;AACzC,IAAA,MAAM,GAAA,GAAM,MAAM,QAAA,CAAS,SAAA,EAAU;AAErC,IAAA,MAAM,UAAA,GAAuG;AAAA,MAC3G;AAAA,KACF;AACA,IAAA,IAAI,GAAA,CAAI,cAAA,EAAgB,UAAA,CAAW,cAAA,GAAiB,GAAA,CAAI,cAAA;AAExD,IAAA,IAAI,GAAA,CAAI,cAAA,EAAgB,UAAA,CAAW,cAAA,GAAiB,GAAA,CAAI,cAAA;AACxD,IAAA,MAAM,MAAA,GAAS,GAAA,CAAI,MAAA,CAAO,UAAU,CAAA;AAEpC,IAAA,IAAI,SAAA,GAAY,KAAA;AAEhB,IAAA,IAAI,SAAA,GAAY,EAAA;AAChB,IAAA,MAAM,YAA6B,EAAC;AAIpC,IAAA,MAAM,gBAAA,GAAmB,CAAC,WAAA,KAAyB;AACjD,MAAA,IAAI,CAAC,cAAA,EAAgB;AACrB,MAAA,MAAM,WAAA,GAAwC,EAAE,GAAG,GAAA,EAAK,MAAA,EAAQ,EAAE,IAAA,EAAM,SAAA,EAAW,SAAA,EAAW,WAAA,EAAY,EAAE;AAC5G,MAAA,OAAA,CAAQ,OAAA,GACL,IAAA,CAAK,MAAM,eAAe,WAAW,CAAC,CAAA,CACtC,KAAA,CAAM,CAAA,KAAA,KAAS;AACd,QAAA,OAAA,CAAQ,IAAA,CAAK,8CAA8C,KAAK,CAAA;AAAA,MAClE,CAAC,CAAA;AAAA,IACL,CAAA;AAEA,IAAA,OAAO,IAAI,cAAA,CAAuB;AAAA,MAChC,KAAA,EAAO,OAAM,UAAA,KAAc;AACzB,QAAA,IAAI,WAAA,GAAc,KAAA;AAClB,QAAA,IAAI;AACF,UAAA,WAAA,MAAiB,KAAA,IAAS,OAAO,UAAA,EAAY;AAC3C,YAAA,IAAI,SAAA,EAAW;AACf,YAAA,IAAI,KAAA,CAAM,SAAS,sBAAA,EAAwB;AAC3C,YAAA,MAAM,UAAU,KAAA,CAAM,OAAA;AACtB,YAAA,IAAI,SAAA,IAAa,OAAA,CAAQ,QAAA,KAAa,SAAA,EAAW;AACjD,YAAA,MAAM,IAAA,GAAO,cAAA,CAAe,OAAA,CAAQ,MAAM,CAAA;AAC1C,YAAA,IAAI,IAAA,EAAM;AACR,cAAA,WAAA,GAAc,IAAA;AACd,cAAA,SAAA,IAAa,IAAA;AACb,cAAA,UAAA,CAAW,QAAQ,IAAI,CAAA;AACvB,cAAA;AAAA,YACF;AAGA,YAAA,MAAM,QAAA,GAAW,kBAAA,CAAmB,OAAA,CAAQ,MAAM,CAAA;AAClD,YAAA,IAAI,QAAA,EAAU;AACZ,cAAA,SAAA,CAAU,KAAK,QAAQ,CAAA;AACvB,cAAA,IAAI,YAAA,EAAc;AAChB,gBAAA,MAAM,MAAA,GAAS,aAAa,QAAQ,CAAA;AACpC,gBAAA,IAAI,MAAA,EAAQ,UAAA,CAAW,OAAA,CAAQ,MAAA,CAAO,QAAA,CAAS,GAAG,CAAA,GAAI,MAAA,GAAS,CAAA,EAAG,MAAM,CAAA,CAAA,CAAG,CAAA;AAAA,cAC7E;AAAA,YACF;AAAA,UACF;AACA,UAAA,IAAI,CAAC,SAAA,IAAa,CAAC,WAAA,IAAe,UAAA,EAAY;AAC5C,YAAA,MAAM,SAAA,GAAY,UAAA,CAAW,MAAM,MAAA,CAAO,MAAM,CAAA;AAChD,YAAA,IAAI,SAAA,EAAW;AACb,cAAA,SAAA,IAAa,SAAA;AACb,cAAA,UAAA,CAAW,QAAQ,SAAS,CAAA;AAAA,YAC9B;AAAA,UACF;AACA,UAAA,IAAI,CAAC,SAAA,EAAW,UAAA,CAAW,KAAA,EAAM;AAEjC,UAAA,gBAAA,CAAiB,SAAS,CAAA;AAAA,QAC5B,SAAS,KAAA,EAAO;AAGd,UAAA,IAAI,SAAA,EAAW;AACb,YAAA,gBAAA,CAAiB,IAAI,CAAA;AACrB,YAAA;AAAA,UACF;AACA,UAAA,UAAA,CAAW,MAAM,KAAK,CAAA;AAAA,QACxB;AAAA,MACF,CAAA;AAAA,MACA,QAAQ,MAAM;AACZ,QAAA,SAAA,GAAY,IAAA;AAGZ,QAAA,KAAK,QAAQ,OAAA,CAAQ,GAAA,CAAI,QAAQ,CAAA,CAAE,MAAM,CAAA,KAAA,KAAS;AAChD,UAAA,OAAA,CAAQ,IAAA,CAAK,kEAAkE,KAAK,CAAA;AAAA,QACtF,CAAC,CAAA;AAAA,MACH;AAAA,KACD,CAAA;AAAA,EACH,CAAA;AACF","file":"chunk-2E3MTAOA.js","sourcesContent":["/** Default LiveKit agent name used for explicit dispatch when none is configured. */\nexport const DEFAULT_LIVEKIT_AGENT_NAME = 'mastra-voice';\n","/**\n * Session metadata passed from the Mastra server to the LiveKit agent worker through\n * LiveKit's job dispatch metadata (a plain string, so this is JSON-serialized).\n */\nexport interface LiveKitSessionMetadata {\n  /** Mastra agent to run, by registered key or agent id. */\n  agentId?: string;\n  /** Memory thread id. Defaults to the LiveKit room name when omitted. */\n  threadId?: string;\n  /** Memory resource id (typically the end user id). */\n  resourceId?: string;\n  /** Plain-object entries restored into a RequestContext for agent execution. */\n  requestContext?: Record<string, unknown>;\n}\n\nexport function parseSessionMetadata(raw: string | undefined | null): LiveKitSessionMetadata {\n  if (!raw) return {};\n  try {\n    const parsed = JSON.parse(raw);\n    if (parsed && typeof parsed === 'object' && !Array.isArray(parsed)) {\n      return parsed as LiveKitSessionMetadata;\n    }\n  } catch {\n    // Dispatch metadata is user-controlled and may not be JSON; treat as absent.\n  }\n  return {};\n}\n\nexport function serializeSessionMetadata(metadata: LiveKitSessionMetadata): string {\n  return JSON.stringify(metadata);\n}\n","import { ReadableStream } from 'node:stream/web';\nimport type { WritableStream } from 'node:stream/web';\nimport type { TracingContext } from '@mastra/core/observability';\nimport type { RequestContext } from '@mastra/core/request-context';\nimport type { Workflow } from '@mastra/core/workflows';\nimport type {\n  VoiceReplyGenerator,\n  VoiceToolCall,\n  VoiceTurnCompleteContext,\n  VoiceTurnCompleteHook,\n  VoiceTurnContext,\n} from './bridge';\n\nexport interface WorkflowReplyGeneratorOptions {\n  /** The Mastra workflow that generates each turn's reply. Runs once per turn (no suspend/resume). */\n  workflow: Workflow;\n  /**\n   * Maps a turn into the workflow's `inputData`. Required — input schemas are caller-defined.\n   * A common shape passes the full transcript so the workflow is stateless between turns, e.g.\n   * `ctx => ({ history: chatContextToMessages(ctx.chatCtx) })`.\n   */\n  workflowInput: (ctx: VoiceTurnContext) => unknown | Promise<unknown>;\n  /**\n   * Only stream text from this step (by id). Defaults to every step that writes text to its\n   * `writer`. Set when multiple steps write and only one produces the spoken reply.\n   */\n  replyStep?: string;\n  /**\n   * Fallback when the workflow streams no text via `writer`: derive the spoken reply from the\n   * final run result. Without this, a non-streaming workflow stays silent. Streaming via the\n   * step `writer` is preferred — it gives the caller low time-to-first-token.\n   */\n  resultText?: (result: unknown) => string | undefined | void;\n  /**\n   * Speak a short phrase while a tool call runs. Fires only for tool calls the reply step surfaces\n   * to its `writer` — use {@link pipeAgentReplyToWriter} (or pipe the agent's `fullStream`) so\n   * `tool-call` chunks reach the stream. See {@link MastraVoiceAgentOptions.toolFeedback}.\n   */\n  toolFeedback?: (toolCall: VoiceToolCall) => string | undefined | void;\n  /**\n   * Fired off the audio path after the reply streams, fire-and-forget. Carries the produced reply\n   * text and any tool calls the workflow surfaced. See {@link MastraVoiceAgentOptions.onTurnComplete}.\n   */\n  onTurnComplete?: VoiceTurnCompleteHook;\n}\n\n/**\n * Unwraps the text carried by a `workflow-step-output` chunk's `payload.output`. A step that\n * pipes `agent.stream().textStream` into its `writer` produces raw strings; a step built from an\n * agent (`createStep(agent)`) produces full `text-delta` chunks. Returns the text for both, or\n * `undefined` for any other shape.\n */\nexport function unwrapStepText(output: unknown): string | undefined {\n  if (typeof output === 'string') return output;\n  if (output && typeof output === 'object' && (output as { type?: unknown }).type === 'text-delta') {\n    const text = (output as { payload?: { text?: unknown } }).payload?.text;\n    return typeof text === 'string' ? text : undefined;\n  }\n  return undefined;\n}\n\n/**\n * Unwraps a tool call carried by a `workflow-step-output` chunk's `payload.output`. A step that\n * pipes the agent's `fullStream` (rather than just `textStream`) into its `writer` surfaces\n * `tool-call` chunks; this returns the {@link VoiceToolCall} for those, or `undefined` otherwise.\n */\nexport function unwrapStepToolCall(output: unknown): VoiceToolCall | undefined {\n  if (!output || typeof output !== 'object' || (output as { type?: unknown }).type !== 'tool-call') return undefined;\n  const payload = (output as { payload?: { toolCallId?: unknown; toolName?: unknown; args?: unknown } }).payload;\n  if (payload && typeof payload.toolCallId === 'string' && typeof payload.toolName === 'string') {\n    return { toolCallId: payload.toolCallId, toolName: payload.toolName, args: payload.args };\n  }\n  return undefined;\n}\n\n/** The minimal shape of a Mastra agent stream consumed by {@link pipeAgentReplyToWriter}. */\nexport interface AgentReplyStreamLike {\n  fullStream: AsyncIterable<unknown>;\n}\n\n/**\n * Streams a Mastra agent's reply into a workflow step's `writer` for the LiveKit workflow\n * entrypoint — the recommended way to drive a turn's reply from a step.\n *\n * It forwards the agent's text deltas (so text-to-speech starts before the full reply is ready)\n * AND its `tool-call` chunks (so {@link WorkflowReplyGeneratorOptions.toolFeedback} fires and\n * {@link WorkflowReplyGeneratorOptions.onTurnComplete}'s `result.toolCalls` is populated). This is\n * the difference from piping only `agent.stream().textStream`, which silently drops tool calls.\n * Other chunk types (reasoning, tool results, lifecycle) are not forwarded, keeping the spoken\n * stream clean. Returns the accumulated reply text for the step to return.\n *\n * Pass the step's `abortSignal` to `agent.stream(...)` so barge-in stops generation promptly.\n *\n * ```ts\n * const generateResponse = createStep({\n *   // ...\n *   execute: async ({ inputData, mastra, writer, abortSignal }) => {\n *     const stream = await mastra.getAgent('callCenter').stream(inputData.turn, { abortSignal });\n *     const reply = await pipeAgentReplyToWriter(stream, writer);\n *     return { reply };\n *   },\n * });\n * ```\n */\nexport function pipeAgentReplyToWriter(\n  agentStream: AgentReplyStreamLike,\n  writer: WritableStream<unknown>,\n): Promise<string> {\n  let text = '';\n  // Re-emit only the chunks the voice path cares about, then pipe through the step writer — which\n  // reuses pipeTo's proven backpressure + close handling rather than driving the writer by hand.\n  const forwarded = new ReadableStream<unknown>({\n    start: async controller => {\n      for await (const chunk of agentStream.fullStream) {\n        const type = (chunk as { type?: unknown })?.type;\n        if (type === 'text-delta') {\n          const delta = (chunk as { payload?: { text?: unknown } }).payload?.text;\n          if (typeof delta === 'string' && delta) {\n            text += delta;\n            controller.enqueue(chunk);\n          }\n        } else if (type === 'tool-call') {\n          controller.enqueue(chunk);\n        }\n      }\n      controller.close();\n    },\n  });\n  return forwarded.pipeTo(writer).then(() => text);\n}\n\n/**\n * A {@link VoiceReplyGenerator} backed by a Mastra workflow. Per turn it starts a fresh run to\n * completion (LiveKit owns the turn boundary, so there is no suspend/resume and no conversation\n * state carried between turns) and streams the text its steps write to their `writer`.\n *\n * A workflow's own stream emits structured step events, not token deltas — text only surfaces\n * when a step pipes it into the injected `writer`, arriving as `workflow-step-output` chunks. The\n * simplest correct reply step calls {@link pipeAgentReplyToWriter}, which forwards both text and\n * tool calls (and passes the step's `abortSignal` through `agent.stream` so barge-in stops\n * generation promptly). Piping only `agent.stream(...).textStream.pipeTo(writer)` works for text\n * but silently drops tool calls, so {@link WorkflowReplyGeneratorOptions.toolFeedback} and\n * {@link WorkflowReplyGeneratorOptions.onTurnComplete}'s `result.toolCalls` stay empty.\n */\nexport function createWorkflowReplyGenerator(options: WorkflowReplyGeneratorOptions): VoiceReplyGenerator {\n  const { workflow, workflowInput, replyStep, resultText, toolFeedback, onTurnComplete } = options;\n  return async ctx => {\n    const inputData = await workflowInput(ctx);\n    const run = await workflow.createRun();\n\n    const streamArgs: { inputData: unknown; tracingContext?: TracingContext; requestContext?: RequestContext } = {\n      inputData,\n    };\n    if (ctx.tracingContext) streamArgs.tracingContext = ctx.tracingContext;\n    // Forward the per-session request context so workflow steps see it, mirroring the agent path.\n    if (ctx.requestContext) streamArgs.requestContext = ctx.requestContext;\n    const output = run.stream(streamArgs);\n\n    let cancelled = false;\n    // Accumulated as the turn streams so the post-turn hook can see what was actually produced.\n    let replyText = '';\n    const toolCalls: VoiceToolCall[] = [];\n\n    // Fire-and-forget after the reply has streamed: off the audio path and not awaited, so it\n    // never delays the next turn. Errors are logged, not thrown. Mirrors createAgentReplyGenerator.\n    const emitTurnComplete = (interrupted: boolean) => {\n      if (!onTurnComplete) return;\n      const completeCtx: VoiceTurnCompleteContext = { ...ctx, result: { text: replyText, toolCalls, interrupted } };\n      Promise.resolve()\n        .then(() => onTurnComplete(completeCtx))\n        .catch(error => {\n          console.warn('@mastra/livekit: onTurnComplete hook threw', error);\n        });\n    };\n\n    return new ReadableStream<string>({\n      start: async controller => {\n        let streamedAny = false;\n        try {\n          for await (const chunk of output.fullStream) {\n            if (cancelled) break;\n            if (chunk.type !== 'workflow-step-output') continue;\n            const payload = chunk.payload as { output?: unknown; stepName?: unknown };\n            if (replyStep && payload.stepName !== replyStep) continue;\n            const text = unwrapStepText(payload.output);\n            if (text) {\n              streamedAny = true;\n              replyText += text;\n              controller.enqueue(text);\n              continue;\n            }\n            // A tool call only surfaces when the step pipes the agent's fullStream; when it does,\n            // mirror the agent path — record it and speak any toolFeedback filler.\n            const toolCall = unwrapStepToolCall(payload.output);\n            if (toolCall) {\n              toolCalls.push(toolCall);\n              if (toolFeedback) {\n                const filler = toolFeedback(toolCall);\n                if (filler) controller.enqueue(filler.endsWith(' ') ? filler : `${filler} `);\n              }\n            }\n          }\n          if (!cancelled && !streamedAny && resultText) {\n            const finalText = resultText(await output.result);\n            if (finalText) {\n              replyText += finalText;\n              controller.enqueue(finalText);\n            }\n          }\n          if (!cancelled) controller.close();\n          // Success, or a clean barge-in break out of the loop: the turn is done either way.\n          emitTurnComplete(cancelled);\n        } catch (error) {\n          // Barge-in cancels the run; that's not a failure — the turn still completed\n          // (interrupted), so the hook still fires for memory reconciliation.\n          if (cancelled) {\n            emitTurnComplete(true);\n            return;\n          }\n          controller.error(error);\n        }\n      },\n      cancel: () => {\n        cancelled = true;\n        // Barge-in fires this synchronously; swallow any rejection from the run cancellation so a\n        // failed cancel can't surface as an unhandled promise rejection.\n        void Promise.resolve(run.cancel()).catch(error => {\n          console.warn('@mastra/livekit: failed to cancel the workflow run on barge-in', error);\n        });\n      },\n    });\n  };\n}\n"]}