feat(barge-in): add default threshold support and drop http transport

.changeset/bargein-default-threshold-drop-http.md

@@ -0,0 +1,5 @@
+---
+'@livekit/agents': patch
+---
+
+Adaptive interruption detection now omits the threshold from `session.create` unless the user explicitly overrides it, letting the gateway apply its fetched default (surfaced via `default_threshold` on `session.created`). The HTTP transport has been dropped — detection always connects over WebSocket and always requires LiveKit credentials, and its base URL now defaults from `LIVEKIT_INFERENCE_URL` instead of `LIVEKIT_REMOTE_EOT_URL`. Inference requests also send an `X-LiveKit-Worker-Token` header when `LIVEKIT_WORKER_TOKEN` is set (hosted agents); a token supplied via the `--worker-token` CLI flag is now re-exported into the environment so forked job subprocesses inherit it and include the header. The `X-LiveKit-Agent-Id` header is now only attached once the room is connected to avoid leaking an unset local-participant SID. The interruption WebSocket is now closed deterministically on stream teardown (including error and cancel paths) instead of only on graceful completion — previously an orphaned socket leaked per session/activity and accumulated for the worker's lifetime. Mid-session threshold/duration changes via `updateOptions` now reconnect the WebSocket in place rather than closing it and letting the next send error the stream — so option changes no longer consume a failover retry (previously enough updates in a session could exhaust the retry budget and stop interruption detection).

agents/src/inference/interruption/_mock_ws.ts

@@ -0,0 +1,58 @@
+// SPDX-FileCopyrightText: 2026 LiveKit, Inc.
+//
+// SPDX-License-Identifier: Apache-2.0
+import { EventEmitter } from 'node:events';
+
+/**
+ * Minimal stand-in for the `ws` WebSocket, used to drive the interruption transport in tests.
+ *
+ * Lives in its own module (rather than inline in each test) so the `vi.mock('ws')` factory can
+ * `await import()` it without a top-level await — the tsup build transpiles test files to CJS,
+ * which does not support top-level await.
+ */
+export class MockWebSocket extends EventEmitter {
+  static OPEN = 1;
+  static instances: MockWebSocket[] = [];
+
+  readyState = 0; // CONNECTING
+  readonly sent: unknown[] = [];
+  terminated = false;
+
+  constructor(
+    public url: string,
+    public opts: unknown,
+  ) {
+    super();
+    MockWebSocket.instances.push(this);
+  }
+
+  send(data: unknown): void {
+    this.sent.push(data);
+  }
+
+  close(): void {
+    this.readyState = 3; // CLOSED
+    this.emit('close', 1000, Buffer.from(''));
+  }
+
+  terminate(): void {
+    this.terminated = true;
+    this.readyState = 3;
+  }
+
+  /** Simulate a successful upgrade. */
+  simulateOpen(): void {
+    this.readyState = MockWebSocket.OPEN;
+    this.emit('open');
+  }
+
+  /** Simulate the server rejecting the upgrade with an HTTP status. */
+  simulateUnexpectedResponse(statusCode: number): void {
+    this.emit('unexpected-response', {}, { statusCode });
+  }
+
+  /** Simulate a server message frame carrying a JSON payload. */
+  simulateMessage(payload: unknown): void {
+    this.emit('message', Buffer.from(JSON.stringify(payload)));
+  }
+}

agents/src/inference/interruption/defaults.ts

@@ -5,9 +5,8 @@ import type { ApiConnectOptions } from './interruption_stream.js';
 import type { InterruptionOptions } from './types.js';
 
 export const MIN_INTERRUPTION_DURATION_IN_S = 0.025 * 2; // 25ms per frame, 2 consecutive frames
-export const THRESHOLD = 0.5;
 export const MAX_AUDIO_DURATION_IN_S = 3.0;
-export const AUDIO_PREFIX_DURATION_IN_S = 0.5;
+export const AUDIO_PREFIX_DURATION_IN_S = 1.0;
 export const DETECTION_INTERVAL_IN_S = 0.1;
 export const REMOTE_INFERENCE_TIMEOUT_IN_S = 0.7;
 export const SAMPLE_RATE = 16000;
@@ -36,12 +35,13 @@ export function intervalForRetry(
 }
 
 // env-derived fields are resolved in the constructor, not at module load.
+// `threshold` is intentionally omitted: when the user does not override it, it stays undefined
+// so the server applies its fetched default.
 export const interruptionOptionDefaults: Omit<
   InterruptionOptions,
-  'baseUrl' | 'useProxy' | 'apiKey' | 'apiSecret'
+  'baseUrl' | 'apiKey' | 'apiSecret' | 'threshold'
 > = {
   sampleRate: SAMPLE_RATE,
-  threshold: THRESHOLD,
   minFrames: Math.ceil(MIN_INTERRUPTION_DURATION_IN_S * FRAMES_PER_SECOND),
   maxAudioDurationInS: MAX_AUDIO_DURATION_IN_S,
   audioPrefixDurationInS: AUDIO_PREFIX_DURATION_IN_S,