@todoforai/voiceloop 0.1.13 → 0.1.14

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/README.md CHANGED
@@ -17,7 +17,7 @@ A zero-dependency JavaScript library that runs the full voice loop — **VAD →
17
17
  - **Real barge-in** — triggers on transcribed *novel words*, not mic energy, so the agent's own voice never cuts it off.
18
18
  - **Self-echo filtering** — 0 self-interruptions with AEC off (Pipecat cut itself 20/30, OpenAI Realtime 17/30 [bench](bench/results/RESULTS.md#scenario-echo-smalltalk--speakermic-coupling-15db--30ms)).
19
19
  - **First audio <1s** — TTS speaks sentence 1 while the LLM writes sentence 2, and the LLM call starts speculatively during your end-of-turn pause.
20
- - **Serialized turns** — rapid-fire turns, tool results, holds and replays can never talk over each other. Locked in by 178 tests.
20
+ - **Serialized turns** — rapid-fire turns, tool results, holds and replays can never talk over each other. Locked in by 184 tests.
21
21
  - **Local-first** — Silero VAD and Piper TTS run as WASM in the tab: free, no cloud round-trip (CDN, then cache).
22
22
 
23
23
  Extra:
@@ -169,7 +169,8 @@ const agent = new VoiceAgent({
169
169
  // toolGate: only for adapters that run tools themselves — see below; ignore it otherwise
170
170
  for await (const delta of myProvider.stream({ history, system, signal })) {
171
171
  yield { text: delta }; // speech text → streamed into TTS
172
- // yield { tool: 'name', args: {...} }; // tool call → fired during playback
172
+ // yield { tool: 'name', args: {...} }; // tool call → the agent runs it during playback;
173
+ // // the NEXT user turn sees its tool_result
173
174
  }
174
175
  },
175
176
  });
@@ -190,13 +191,13 @@ const agent = new VoiceAgent({
190
191
  });
191
192
  ```
192
193
 
193
- Tool calls execute while the agent is still talking. Results are recorded in a per-turn ledger so the model never re-fires the same call, and `agent.notify('[TOOL RESULT get_weather] 22°C sunny')` relays an async outcome back for a spoken follow-up — bursts of results collapse into one reply instead of three interrupting monologues.
194
+ Tool calls execute while the agent is still talking. The default adapter is a loop: it feeds each result back to the model inside the same reply, so the model keeps talking with the answer in hand (bounded to 6 model messages per turn). Results are recorded natively (tool_use / tool_result) so the model never re-fires the same call, and `agent.notify('[TOOL RESULT get_weather] 22°C sunny')` relays an async outcome back for a spoken follow-up — bursts of results collapse into one reply instead of three interrupting monologues.
194
195
 
195
196
  Tools are **dispatched, not awaited**: the turn completes when the agent stops speaking, whatever the tool is still doing. A slow or hung tool can never stall the conversation, and a tool is free to `await agent.notify(...)` with its own result. Long work belongs in a tool that returns promptly ("checking that now") and delivers the outcome later via `notify()`.
196
197
 
197
198
  ### Adapters that run tools themselves
198
199
 
199
- Some adapters (agent loops) execute tools **inside** the generator, feeding results back to the model so it keeps talking with the answer in hand — all within one turn. Declare two flags so voiceloop adapts:
200
+ The default `makeOpenAILLM` and any agent loop execute tools **inside** the generator, feeding results back to the model so it keeps talking with the answer in hand — all within one turn. A custom loop declares two flags so voiceloop adapts, and yields `{ newMessage: true }` where the model's next message starts:
200
201
 
201
202
  ```js
202
203
  async function* myAgentLoop(history, system, signal, { toolGate } = {}) {
@@ -204,6 +205,7 @@ async function* myAgentLoop(history, system, signal, { toolGate } = {}) {
204
205
  if (ev.type === 'text') yield { text: ev.delta };
205
206
  if (ev.type === 'tool_start') yield { tool: ev.name, id: ev.id, args: ev.args, running: true };
206
207
  if (ev.type === 'tool_end') yield { tool: ev.name, id: ev.id, args: ev.args, result: ev.result };
208
+ if (ev.type === 'message_start' && !first) yield { newMessage: true };
207
209
  }
208
210
  }
209
211
  myAgentLoop.executesTools = true; // chunks carry the outcome — voiceloop reports, never re-runs
package/package.json CHANGED
@@ -1,6 +1,6 @@
1
1
  {
2
2
  "name": "@todoforai/voiceloop",
3
- "version": "0.1.13",
3
+ "version": "0.1.14",
4
4
  "description": "Fast, interruptible browser voice agent: VAD → STT → LLM → TTS with real barge-in, echo filtering, streaming first-sentence TTS and tap-to-seek. Zero runtime dependencies.",
5
5
  "type": "module",
6
6
  "main": "src/index.js",
package/src/index.d.ts CHANGED
@@ -59,9 +59,10 @@ export type HistoryBlock =
59
59
  | { type: 'tool_result'; tool_use_id: string; content: string; is_error?: boolean };
60
60
  export type HistoryMessage = { role: string; content: string | HistoryBlock[] };
61
61
  // Pluggable LLM: lazy async generator over the chat history (fetch fires on first pull).
62
- // The default (makeOpenAILLM) yields bare tool chunks that the AGENT executes; their results are
63
- // recorded as tool_result and the agent starts a follow-up turn so the model reacts to them.
64
- // An agent LOOP that runs tools inside its own generator declares the two flags below.
62
+ // The default (makeOpenAILLM) is a tool LOOP: it runs each call inside the generator, feeds the
63
+ // result back to the model and marks the model's next message with { newMessage }. A plain adapter
64
+ // (no flags) yields bare tool chunks that the AGENT executes and records; the NEXT user turn sees
65
+ // the tool_result (no reaction turn is started).
65
66
  export interface LLM {
66
67
  (
67
68
  history: HistoryMessage[],
@@ -71,12 +72,12 @@ export interface LLM {
71
72
  ): AsyncIterable<LLMChunk>;
72
73
  /** This generator is a tool LOOP: it runs every call itself and feeds the result back inside the
73
74
  * same reply (chunks carry `result`/`error`, optionally preceded by a `running` announcement).
74
- * The agent never executes its calls and never starts a reaction turn for them. */
75
+ * The agent never executes its calls. */
75
76
  executesTools?: boolean;
76
77
  /** This generator awaits `options.toolGate` before ANY tool executes. Required to keep
77
78
  * speculative prefetch on: prefetch starts from an uncommitted interim, and the gate is what
78
79
  * stops a speculation from firing a real side effect. `executesTools` without this silently
79
- * disables prefetch (see voice-agent.js:793). */
80
+ * disables prefetch (see _startPrefetch in voice-agent.js). */
80
81
  acceptsToolGate?: boolean;
81
82
  }
82
83
 
package/src/llm-openai.js CHANGED
@@ -1,9 +1,10 @@
1
1
  // ── Default LLM: any OpenAI-compatible /chat/completions endpoint ───────────────────────────────
2
- // Normalizes the OpenAI streaming wire format to the VoiceAgent chunk contract:
3
- // { text } — one per content delta (streamed straight into the TTS)
4
- // { tool, args, id }— one per COMPLETED tool call (arguments accumulate across deltas; a call
5
- // whose accumulated arguments never parse as JSON is DROPPED with a warning —
6
- // firing a side-effecting tool with guessed/empty args would be worse)
2
+ // A LOOP adapter (executesTools + acceptsToolGate): one turn may span several model messages. Each
3
+ // completion streams { text } deltas straight to the TTS; a COMPLETED tool call (arguments accumulate
4
+ // across deltas; a call whose arguments never parse as JSON is DROPPED with a warning — firing a
5
+ // side-effecting tool with guessed args would be worse) is announced ({ running: true }), executed
6
+ // here, reported with its outcome, fed back to the model as a tool message, and the model's next
7
+ // message starts after a { newMessage } boundary — bounded by MAX_ROUNDS per turn.
7
8
  //
8
9
  // Works against OpenAI, Groq, Cerebras, Together, OpenRouter, Ollama, vLLM, LiteLLM… anything that
9
10
  // speaks /chat/completions SSE. NEVER ship a provider secret key to a public page — in production
@@ -37,6 +38,19 @@ export const toOpenAIMessages = (history) => history.flatMap((m) => {
37
38
  return text ? [...results, { role: m.role, content: text }] : results; // OpenAI has no is_error → text-marked
38
39
  });
39
40
 
41
+ // Render a host tool's return value as model-facing text (the tool_result content in history, and
42
+ // what adapters feed back to the model). Never throws: the side effect already RAN, so an
43
+ // unserializable result must not retro-fail the call and invite a retry.
44
+ export const toolResultText = (result) => {
45
+ if (result === undefined) return 'ok';
46
+ if (typeof result === 'string') return result;
47
+ if (result && typeof result.text === 'string' && typeof result.ok === 'boolean')
48
+ return result.ok ? result.text : `[failed] ${result.text}`;
49
+ try { return JSON.stringify(result) ?? 'ok'; } catch { return String(result); }
50
+ };
51
+
52
+ const MAX_ROUNDS = 6; // model messages per user turn — a model that keeps calling tools then waits for the user
53
+
40
54
  export function makeOpenAILLM({ llmUrl, apiKey = '', model = '', maxTokens = 1024, tools = {}, fetchFn, extraBody = {} } = {}) {
41
55
  const toolSpecs = Object.entries(tools).map(([name, t]) => ({
42
56
  type: 'function',
@@ -47,15 +61,15 @@ export function makeOpenAILLM({ llmUrl, apiKey = '', model = '', maxTokens = 102
47
61
  },
48
62
  }));
49
63
 
50
- return async function* (history, system, signal) {
51
- if (!llmUrl) throw new Error('VoiceAgent: no `llm` generator and no `llmUrl` configured — pass one of them');
64
+ // One completion: streams { text } and COMPLETED { tool, args, id } chunks for a wire-shaped message list.
65
+ async function* streamOnce(messages, signal) {
52
66
  const doFetch = fetchFn ?? fetch;
53
67
  const r = await doFetch(llmUrl, {
54
68
  method: 'POST', signal,
55
69
  headers: { 'Content-Type': 'application/json', ...(apiKey ? { Authorization: `Bearer ${apiKey}` } : {}) },
56
70
  body: JSON.stringify({
57
71
  stream: true,
58
- messages: [{ role: 'system', content: system }, ...toOpenAIMessages(history)],
72
+ messages,
59
73
  max_tokens: maxTokens,
60
74
  ...(model ? { model } : {}),
61
75
  ...(toolSpecs.length ? { tools: toolSpecs } : {}),
@@ -76,9 +90,6 @@ export function makeOpenAILLM({ llmUrl, apiKey = '', model = '', maxTokens = 102
76
90
  for (const c of pending.values()) {
77
91
  if (!c.name) continue;
78
92
  let args;
79
- // Truncated/invalid argument JSON (aborted stream, provider bug) → DROP the call. The agent
80
- // dispatches tools immediately, so guessing `{}` could fire a real side effect with the
81
- // wrong arguments.
82
93
  try { args = c.args ? JSON.parse(c.args) : {}; }
83
94
  catch { console.warn(`makeOpenAILLM: dropping tool call "${c.name}" — arguments never parsed: ${c.args}`); continue; }
84
95
  yield { tool: c.name, args, ...(c.id ? { id: c.id } : {}) };
@@ -86,9 +97,8 @@ export function makeOpenAILLM({ llmUrl, apiKey = '', model = '', maxTokens = 102
86
97
  pending.clear();
87
98
  };
88
99
 
89
- // One SSE data payload → zero or more chunks. Returns 'done' on [DONE].
100
+ // One SSE data payload → zero or more chunks.
90
101
  const handle = function* (payload) {
91
- if (payload === '[DONE]') return; // caller checks the sentinel itself
92
102
  let m; try { m = JSON.parse(payload); } catch { return; } // tolerate keep-alive junk
93
103
  if (m.error) throw new Error(m.error.message ?? String(m.error));
94
104
  const choice = m.choices?.[0];
@@ -136,5 +146,46 @@ export function makeOpenAILLM({ llmUrl, apiKey = '', model = '', maxTokens = 102
136
146
  } finally {
137
147
  reader.cancel().catch(() => {});
138
148
  }
149
+ }
150
+
151
+ const gen = async function* (history, system, signal, { toolGate } = {}) {
152
+ if (!llmUrl) throw new Error('VoiceAgent: no `llm` generator and no `llmUrl` configured — pass one of them');
153
+ // Anthropic-shaped working history (what VoiceAgent keeps); re-serialized to the wire every round
154
+ // so tool_use/tool_result pairing stays in ONE place (toOpenAIMessages).
155
+ const hist = [...history];
156
+ for (let round = 0; round < MAX_ROUNDS; round++) {
157
+ let text = '';
158
+ const calls = []; // { id, tool, args, outcome: Promise<{result}|{error}> }
159
+ const seen = new Set();
160
+ for await (const c of streamOnce([{ role: 'system', content: system }, ...toOpenAIMessages(hist)], signal)) {
161
+ if (c.text) { text += c.text; yield c; continue; }
162
+ const id = c.id || `call-${round}-${calls.length + 1}`; // announcements need an id to be replaced in place
163
+ if (seen.has(id)) continue;
164
+ seen.add(id);
165
+ yield { tool: c.tool, id, args: c.args, running: true };
166
+ // Execute NOW, in parallel with TTS. A speculative run waits at the gate: it rejects when the
167
+ // speculation is dropped, so the tool never fires for a turn the user was still amending.
168
+ const outcome = (async () => {
169
+ if (toolGate) await toolGate;
170
+ try { return { result: await tools[c.tool]?.run?.(c.args) }; }
171
+ catch (e) { return { error: e?.message || String(e ?? 'failed') }; }
172
+ })();
173
+ calls.push({ id, tool: c.tool, args: c.args, outcome });
174
+ }
175
+ if (!calls.length) return;
176
+ const results = [];
177
+ for (const c of calls) {
178
+ const o = await c.outcome;
179
+ yield { tool: c.tool, id: c.id, args: c.args, ...o };
180
+ const failed = 'error' in o || (o.result && typeof o.result === 'object' && o.result.ok === false);
181
+ results.push({ type: 'tool_result', tool_use_id: c.id, content: 'error' in o ? String(o.error) : toolResultText(o.result), ...(failed ? { is_error: true } : {}) });
182
+ }
183
+ hist.push({ role: 'assistant', content: [...(text ? [{ type: 'text', text }] : []), ...calls.map((c) => ({ type: 'tool_use', id: c.id, name: c.tool, input: c.args ?? {} }))] });
184
+ hist.push({ role: 'user', content: results });
185
+ yield { newMessage: true };
186
+ }
139
187
  };
188
+ gen.executesTools = true;
189
+ gen.acceptsToolGate = true;
190
+ return gen;
140
191
  }
@@ -19,7 +19,8 @@
19
19
 
20
20
  import { STT_PROVIDERS, resolveSttProvider } from './stt.js';
21
21
  import { TUNING } from './tuning.js';
22
- import { makeOpenAILLM } from './llm-openai.js';
22
+ import { makeOpenAILLM, toolResultText } from './llm-openai.js';
23
+ export { toolResultText };
23
24
 
24
25
  // Default persona: a minimal voice-first system prompt. Hosts replace it via `persona`
25
26
  // (who the agent is) and append live context via `sysmsg` / setSysmsg().
@@ -79,21 +80,10 @@ const withTimeout = (p, ms, fallback) => {
79
80
  const callArgs = (c) => JSON.stringify(c.args ?? {}, (_k, v) =>
80
81
  v && typeof v === 'object' && !Array.isArray(v) ? Object.fromEntries(Object.keys(v).sort().map((k) => [k, v[k]])) : v);
81
82
  const callKey = (c) => c.id ? `id:${c.id}` : `${c.tool} ${callArgs(c)}`;
82
- // Render a host tool's return value as model-facing text (the tool_result content in history, and
83
- // what LLM adapters feed back to the model). Never throws: the side effect already RAN,
84
- // so an unserializable result must not retro-fail the call and invite a retry.
85
- export const toolResultText = (result) => {
86
- if (result === undefined) return 'ok';
87
- if (typeof result === 'string') return result;
88
- if (result && typeof result.text === 'string' && typeof result.ok === 'boolean')
89
- return result.ok ? result.text : `[failed] ${result.text}`;
90
- try { return JSON.stringify(result) ?? 'ok'; } catch { return String(result); }
91
- };
92
83
  // One model-facing history record for a settled/pending tool call: the native tool_result block that
93
84
  // pairs with its tool_use. Bounded — history is spoken-conversation-sized, not a log. A call whose
94
85
  // outcome has not landed yet says so; _runTool patches the SAME block in place when it does.
95
86
  const RESULT_MAX = 4000;
96
- const MAX_TOOL_ROUNDS = 6;
97
87
  const PENDING_RESULT = 'still running — the result will follow in a later turn';
98
88
  const bound = (text) => text.length > RESULT_MAX ? `${text.slice(0, RESULT_MAX)}… (${text.length} chars)` : text;
99
89
  const toolResultBlock = (c) => {
@@ -314,8 +304,6 @@ export class VoiceAgent {
314
304
  this.state = 'idle'; // idle | listening | thinking | speaking
315
305
  this._abort = null; // aborts the in-flight LLM
316
306
  this._replySeq = 0; // bumped per reply-generating turn; a queued turn drops itself if a newer one exists
317
- this._toolRounds = 0; // tool-result reactions since the user last spoke (see _runTurn react)
318
- this._unseenResults = false; // a tool_result was patched after the last request was built (see _runTurn react)
319
307
  this._preroll = []; this._wasSpeaking = false; this._closed = false; this._destroyed = false;
320
308
  // Diagnostic tap: last ~30s of the exact i16 frames handed to STT + delivery-stall gaps, for
321
309
  // dumpAudio() (replay any "it misheard me" run and prove capture clean-or-not).
@@ -637,7 +625,6 @@ export class VoiceAgent {
637
625
  // history stays ordered (…, assistant-so-far, user, user). Held pushes NEVER start an LLM turn —
638
626
  // release (_flush) owns starting the single turn, and waits for all of these to settle first.
639
627
  _pushHeld(text) {
640
- this._toolRounds = 0;
641
628
  const p = (async () => { await this._holdBarrier; this.history.push({ role: 'user', content: text }); })();
642
629
  this._heldPushes.add(p);
643
630
  p.finally(() => this._heldPushes.delete(p));
@@ -700,7 +687,6 @@ export class VoiceAgent {
700
687
  // caller (errors surface via onEvent), so it's safe to call without awaiting/catching.
701
688
  _onUserTurn(text, { speech = false } = {}) {
702
689
  if (!text.trim()) return Promise.resolve();
703
- this._toolRounds = 0;
704
690
  return this._runTurn(text, { speech });
705
691
  }
706
692
 
@@ -737,8 +723,8 @@ export class VoiceAgent {
737
723
  _startPrefetch(text) {
738
724
  // Speculating is only safe if pulling the first chunk cannot fire a real side effect on a
739
725
  // transcript the user is still speaking and may yet change — abort cannot undo a sent email.
740
- // The default LLM's chunks are inert (tools run later, in _speakTurn). An LLM that executes
741
- // tools INSIDE its generator (llm.executesTools) is safe only if it also honors `toolGate`
726
+ // An LLM whose chunks are inert (no tool execution) is always safe. One that executes tools
727
+ // INSIDE its generator (llm.executesTools) is safe only if it also honors `toolGate`
742
728
  // (llm.acceptsToolGate): we hand it a promise it must await before ANY tool runs, so text
743
729
  // streams speculatively while tools wait for the turn to commit. Neither → don't speculate.
744
730
  if (this.llm.executesTools && !this.llm.acceptsToolGate) return;
@@ -817,7 +803,7 @@ export class VoiceAgent {
817
803
  // if nothing newer has queued behind it, and never cut the reply in flight. Anything newer already
818
804
  // sees this message in history and answers over the fuller picture — so a burst of tool results
819
805
  // collapses into ONE reply instead of one interrupting monologue per result.
820
- _runTurn(text, { supersede = true, speech = false, react = false } = {}) {
806
+ _runTurn(text, { supersede = true, speech = false } = {}) {
821
807
  const prev = this._turn;
822
808
  if (supersede) {
823
809
  this._abort?.abort(); // cut the in-flight reply NOW (barge-in / superseded)
@@ -843,24 +829,9 @@ export class VoiceAgent {
843
829
  // Only a SPOKEN final may adopt (it seeded the speculation); typed/notify/held turns drop it.
844
830
  const pf = speech && text ? this._takePrefetch(text) : (this._dropPrefetch(), null);
845
831
  if (text) this.history.push({ role: 'user', content: text });
846
- // A reaction turn owes a reply only for results no request has carried yet: either the
847
- // tool_result record is still the last message, or a late outcome was patched in after the
848
- // last request was built (_unseenResults; cleared when a request starts) — then a short
849
- // [EVENT] user turn, the host's channel for out-of-band facts, points the model at the patched
850
- // block (a request can't end on an assistant message). Bounded per user utterance
851
- // (_toolRounds, reset on every accepted user turn): a model that keeps calling tools gets
852
- // MAX_TOOL_ROUNDS reactions, then waits for the user.
853
- if (react) {
854
- const last = this.history[this.history.length - 1];
855
- const owed = (last?.role === 'user' && Array.isArray(last.content)) || this._unseenResults;
856
- if (!owed || this._toolRounds >= MAX_TOOL_ROUNDS) { this._unseenResults = false; return; }
857
- this._toolRounds++;
858
- if (last?.role !== 'user') this.history.push({ role: 'user', content: '[EVENT tool_result] a pending tool finished — see its tool_result above.' });
859
- }
860
832
  // Reply only if the conversation now ends on a user turn (a raced turn may have already
861
833
  // replied). pf is always null here: if we pushed a user turn it IS the last entry.
862
834
  if (this.history[this.history.length - 1]?.role !== 'user') return;
863
- this._unseenResults = false; // this request carries every result landed so far
864
835
  this._set('thinking'); // LLM stage
865
836
  this._abort = pf?.ctl ?? new AbortController(); // adopted prefetch keeps ITS controller so barge-in aborts the right fetch
866
837
  await this._speakTurn(this._abort.signal, pf?.gen, seq); // _speakTurn never throws — errors surface via onEvent
@@ -888,9 +859,9 @@ export class VoiceAgent {
888
859
  // notation it could imitate. `this.llm(...)` is a lazy generator — its fetch fires only when
889
860
  // tts.speak first pulls, by which point any prior turn is aborted; keeps LLM streams serial.
890
861
  //
891
- // Round boundary: the default adapter's stream is ONE model message (text + calls together;
892
- // the reaction to the results comes from _continue). A loop adapter feeds results back inside
893
- // the stream and marks where the model's NEXT message starts with a { newMessage } chunk.
862
+ // Round boundary: a loop adapter (executesTools) feeds results back inside the stream and marks
863
+ // where the model's NEXT message starts with a { newMessage } chunk. A plain adapter's stream is
864
+ // ONE model message; the agent runs its calls and the NEXT user turn sees the results.
894
865
  let answer = '';
895
866
  const rounds = [], calls = []; // calls: every act this turn saw, for dedup — outlives recording
896
867
  const loop = !!this.llm.executesTools;
@@ -898,27 +869,23 @@ export class VoiceAgent {
898
869
  const newRound = () => { const r = rounds[rounds.length - 1]; if (r && (r.text || r.calls.length)) rounds.push({ text: '', calls: [] }); };
899
870
  let n = 0;
900
871
  // Accept a call once: identity is the provider id, else name+args (an id-less repeat is a
901
- // provider re-emitting its buffered call). A loop adapter owns execution AND feeds the result
902
- // back inside its reply; anything else is agent-run (or agent-reported when the chunk already
903
- // carries an outcome) and gets a reaction turn.
872
+ // provider re-emitting its buffered call).
904
873
  const accept = (item) => {
905
874
  const key = callKey(item);
906
875
  let c = calls.find((x) => x.key === key);
907
876
  if (c) return c;
908
- c = { ...item, key, id: item.id || `call-${turn}-${++n}`, owner: loop ? 'adapter' : 'agent' };
877
+ c = { ...item, key, id: item.id || `call-${turn}-${++n}` };
909
878
  calls.push(c); round().calls.push(c);
910
879
  return c;
911
880
  };
912
881
  const outcome = (c, item) => { if ('result' in item) c.result = item.result; else if ('error' in item) c.error = item.error; };
913
882
  const hasOutcome = (c) => 'result' in c || 'error' in c;
914
883
  // An outcome may land AFTER the turn recorded the call (a hung tool, a barge-in mid-execution):
915
- // its tool_result block is patched IN PLACE so the next request carries the real result, and the
916
- // model gets to react to it (see _continue).
884
+ // its tool_result block is patched IN PLACE so the next request carries the real result.
917
885
  const settle = (c) => this._runTool(c).then(() => {
918
886
  if (!c.block) return; // not recorded yet — record() picks the outcome up
919
887
  if (!c.consumed) { Object.assign(c.block, toolResultBlock(c)); c.consumed = true; }
920
888
  this._dropPrefetch(); // a speculation serialized the pending block — stale
921
- this._unseenResults = true; this._continue();
922
889
  });
923
890
  const record = (keptLen) => {
924
891
  let budget = keptLen;
@@ -936,11 +903,6 @@ export class VoiceAgent {
936
903
  }
937
904
  rounds.length = 0;
938
905
  };
939
- // Follow-up only for calls the AGENT ran: a loop adapter fed its results back inside the reply.
940
- const recordAndContinue = (keptLen) => {
941
- record(keptLen);
942
- if (calls.some((c) => c.owner === 'agent' && c.consumed)) this._continue();
943
- };
944
906
  // Pre-execution markers: an LLM that runs tools inside its own generator (executesTools) can
945
907
  // announce a call with `running: true` BEFORE it executes, so the host shows a live spinner chip
946
908
  // that the outcome chunk replaces in place (same id). A turn that dies before an announced call's
@@ -979,7 +941,7 @@ export class VoiceAgent {
979
941
  if (finalized && !hasOutcome(item)) continue; // late bare call: the user cut this reply off
980
942
  const c = accept(item);
981
943
  outcome(c, item);
982
- if (finalized) record(0); // late outcome of an unseen call: give it a record of its own (settle schedules the reaction)
944
+ if (finalized) record(0); // late outcome of an unseen call: give it a record of its own
983
945
  settle(c);
984
946
  continue;
985
947
  }
@@ -1043,7 +1005,7 @@ export class VoiceAgent {
1043
1005
  // Tools fire DURING streaming, so a reply that dies here may already have dispatched one.
1044
1006
  // Record what was done and heard so far — dropping it would let the next turn call the
1045
1007
  // same tool again.
1046
- recordAndContinue(signal.keepFull ? answer.length : this._replyText.length);
1008
+ record(signal.keepFull ? answer.length : this._replyText.length);
1047
1009
  finishInterrupted(); // announced-but-unfinished → failed chip, not a stuck spinner
1048
1010
  if (this.state !== 'idle') this._set('listening');
1049
1011
  return;
@@ -1067,7 +1029,7 @@ export class VoiceAgent {
1067
1029
  // tail aligns). keepFull/normal-finish → whole answer (no tail); a barge-in → the heard prefix
1068
1030
  // solid + the unspoken remainder dimmed, showing exactly where TTS got.
1069
1031
  const kept = signal.keepFull ? answer : rawPrefix(spoken);
1070
- recordAndContinue(kept.length);
1032
+ record(kept.length);
1071
1033
  if (kept) {
1072
1034
  this.onEvent({ type: 'assistant', text: kept, full: answer, final: true, turn });
1073
1035
  } else if (answer) {
@@ -1085,19 +1047,6 @@ export class VoiceAgent {
1085
1047
  if (!signal.aborted && this.state !== 'idle') this._set('listening'); // superseded turns don't touch state
1086
1048
  }
1087
1049
 
1088
- // Native tool semantics: once a tool_result is in history the model gets to react to it. Queues a
1089
- // SOFT reaction turn (behind the running one; dropped if a newer user turn supersedes it — that
1090
- // turn sees the result anyway) after the agent ran a call for the default adapter (its stream
1091
- // ends at the call; the result was never fed back), or when a LATE outcome (hung/slow tool,
1092
- // barge-in mid-execution) landed in a record. Whether a reply is still owed is decided when the
1093
- // turn RUNS (see _runTurn `react`), since the predecessor may already have carried the result.
1094
- // Loop adapters fed the result back inside the reply, so nothing fires for them. Silent while
1095
- // the user is speaking — their utterance will see the result in history anyway.
1096
- _continue() {
1097
- if (this._closed || this._speaking) return;
1098
- this._runTurn('', { supersede: false, react: true });
1099
- }
1100
-
1101
1050
  // Run one chosen tool call and report it. Fired as soon as its tool_use arrives in the stream, in
1102
1051
  // parallel with TTS, and NEVER joined: the turn records the call (tool_result pending) and finishes
1103
1052
  // when the speaking does (see _speakTurn), so a tool that hangs costs its own result and nothing