@openclaw/voice-call 2026.8.1-beta.3 → 2026.8.2

package/dist/{runtime-entry-DK8kPCGC.js → runtime-entry-CJkbbEgg.js}

@@ -1,7 +1,7 @@
 import { TtsConfigSchema, isBlockedHostnameOrIp, isRequestBodyLimitError, readRequestBodyWithLimit, requestBodyErrorToText } from "./runtime-api.js";
 import "./api.js";
 import { t as TWILIO_REGIONS } from "./twilio-region-4fkgz3UG.js";
-import { _ as releaseRejectedProviderCall, b as setVoiceCallStateRuntime, d as loadActiveCallsFromStore, g as appendCallReplayKey, l as findCallMatchesInStore, p as persistCallRecord, t as resolveDefaultVoiceCallStoreDir, u as getCallHistoryFromStore, v as rememberManagerReplayKey, x as TerminalStates, y as reserveRejectedProviderCall } from "./store-path-51_oC_xn.js";
+import { _ as releaseRejectedProviderCall, b as setVoiceCallStateRuntime, d as loadActiveCallsFromStore, g as appendCallReplayKey, l as findCallInStore, p as persistCallRecord, t as resolveDefaultVoiceCallStoreDir, u as getCallHistoryFromStore, v as rememberManagerReplayKey, x as TerminalStates, y as reserveRejectedProviderCall } from "./store-path-C_z9zKnA.js";
 import { formatErrorMessage } from "openclaw/plugin-sdk/error-runtime";
 import { isLoopbackHost } from "openclaw/plugin-sdk/gateway-runtime";
 import { normalizeAgentId, parseAgentSessionKey } from "openclaw/plugin-sdk/routing";
@@ -1077,8 +1077,8 @@ async function initiateCall(ctx, to, sessionKey, options) {
 			...requesterSessionKey ? { requesterSessionKey } : {}
 		}
 	};
-	ctx.activeCalls.set(callId, callRecord);
 	persistCallRecord(ctx.storePath, callRecord);
+	ctx.activeCalls.set(callId, callRecord);
 	try {
 		let inlineTwiml;
 		let preConnectTwiml;
@@ -1109,10 +1109,12 @@ async function initiateCall(ctx, to, sessionKey, options) {
 				streamAuthToken: streamSession.token
 			} : {}
 		});
-		callRecord.providerCallId = result.providerCallId;
-		ctx.providerCallIdMap.set(result.providerCallId, callId);
-		persistCallRecord(ctx.storePath, callRecord);
-		console.log(`[voice-call] Outbound call initiated: callId=${callId} providerCallId=${result.providerCallId} mode=${mode} preConnectDtmf=${preConnectTwiml ? "yes" : "no"} initialMessage=${initialMessage ? "yes" : "no"}`);
+		if (ctx.activeCalls.get(callId) === callRecord && !callRecord.providerCallId) {
+			callRecord.providerCallId = result.providerCallId;
+			ctx.providerCallIdMap.set(result.providerCallId, callId);
+			persistCallRecord(ctx.storePath, callRecord);
+		}
+		console.log(`[voice-call] Outbound call initiated: callId=${callId} providerCallId=${callRecord.providerCallId ?? result.providerCallId} mode=${mode} preConnectDtmf=${preConnectTwiml ? "yes" : "no"} initialMessage=${initialMessage ? "yes" : "no"}`);
 		return {
 			callId,
 			success: true
@@ -1442,7 +1444,22 @@ function processEvent(ctx, event) {
 		providerCallIdMap: ctx.providerCallIdMap,
 		callIdOrProviderCallId: event.callId
 	});
-	const providerCallId = event.providerCallId;
+	let providerCallId = event.providerCallId;
+	let retained;
+	if (!call) {
+		retained = findCallInStore(ctx.storePath, event.callId);
+		if (!retained && providerCallId && providerCallId !== event.callId) retained = findCallInStore(ctx.storePath, providerCallId);
+		if (retained && retained.metadata?.rejectionReason !== "inbound-policy") {
+			call = ctx.activeCalls.get(retained.callId);
+			if (!call) return TerminalStates.has(retained.state) ? { kind: "ignored" } : {
+				kind: "ignored",
+				replayable: true
+			};
+		}
+	}
+	if (call && providerCallId && providerCallId !== call.providerCallId) {
+		if ((providerCallId === event.callId && retained ? retained : findCallInStore(ctx.storePath, providerCallId))?.callId === call.callId) providerCallId = call.providerCallId;
+	}
 	const eventDirection = event.direction === "inbound" || event.direction === "outbound" ? event.direction : void 0;
 	if (!call && providerCallId && eventDirection) {
 		if (eventDirection === "inbound" && !shouldAcceptInbound(ctx.config, event.from)) {
@@ -1514,14 +1531,14 @@ function processEvent(ctx, event) {
 		}
 	};
 	const publishProviderCallId = (terminal = false) => {
-		if (!event.providerCallId || event.providerCallId === previousCall.providerCallId) return;
-		if (!terminal) ctx.providerCallIdMap.set(event.providerCallId, activeCall.callId);
+		if (!providerCallId || providerCallId === previousCall.providerCallId) return;
+		if (!terminal) ctx.providerCallIdMap.set(providerCallId, activeCall.callId);
 		if (previousCall.providerCallId) {
 			if (ctx.providerCallIdMap.get(previousCall.providerCallId) === activeCall.callId) ctx.providerCallIdMap.delete(previousCall.providerCallId);
 		}
 	};
 	try {
-		if (event.providerCallId && event.providerCallId !== activeCall.providerCallId) activeCall.providerCallId = event.providerCallId;
+		if (providerCallId && providerCallId !== activeCall.providerCallId) activeCall.providerCallId = providerCallId;
 		if (shouldCommitReplayKey) appendCallReplayKey(activeCall.processedEventIds, dedupeKey);
 		switch (event.type) {
 			case "call.initiated": {
@@ -1587,6 +1604,7 @@ function processEvent(ctx, event) {
 						resolveTranscriptWaiter(ctx, activeCall.callId, event.transcript, event.turnToken);
 					});
 				}
+				if (event.transcript.trim()) effects.push(() => ctx.onCallerSpeech?.(activeCall));
 				prepareLiveDurationTimer();
 				transitionState(activeCall, "listening");
 				break;
@@ -1619,7 +1637,6 @@ function processEvent(ctx, event) {
 					kind: "processed",
 					replayable: true
 				};
-				break;
 		}
 		persistCallRecord(ctx.storePath, activeCall);
 		publishProviderCallId();
@@ -1646,7 +1663,7 @@ function resolveUserPath(input) {
 	const trimmed = input.trim();
 	if (!trimmed) return trimmed;
 	if (trimmed.startsWith("~")) {
-		const expanded = trimmed.replace(/^~(?=$|[\\/])/, os.homedir());
+		const expanded = trimmed.replace(/^~(?=$|[\\/])/, () => os.homedir());
 		return path.resolve(expanded);
 	}
 	return path.resolve(trimmed);
@@ -1686,6 +1703,7 @@ var CallManager = class {
 		this.transcriptWaiters = /* @__PURE__ */ new Map();
 		this.maxDurationTimers = /* @__PURE__ */ new Map();
 		this.initialMessageInFlight = /* @__PURE__ */ new Set();
+		this.autoResponseOwners = /* @__PURE__ */ new WeakMap();
 		this.config = config;
 		this.coreSession = coreSession;
 		this.storePath = resolveDefaultStoreBase(config, storePath);
@@ -1857,6 +1875,7 @@ var CallManager = class {
 			transcriptWaiters: this.transcriptWaiters,
 			maxDurationTimers: this.maxDurationTimers,
 			initialMessageInFlight: this.initialMessageInFlight,
+			onCallerSpeech: (call) => this.invalidateAutoResponse(call),
 			onCallAnswered: (call) => {
 				this.maybeSpeakInitialMessageOnAnswered(call);
 			},
@@ -1869,6 +1888,19 @@ var CallManager = class {
 	processEvent(event) {
 		return processEvent(this.getContext(), event);
 	}
+	createAutoResponseGuard(call) {
+		const owner = Symbol("automatic response");
+		this.autoResponseOwners.set(call, owner);
+		return {
+			isCurrent: () => this.activeCalls.get(call.callId) === call && !TerminalStates.has(call.state) && this.autoResponseOwners.get(call) === owner,
+			release: () => {
+				if (this.autoResponseOwners.get(call) === owner) this.autoResponseOwners.delete(call);
+			}
+		};
+	}
+	invalidateAutoResponse(call) {
+		this.autoResponseOwners.delete(call);
+	}
 	shouldDeferConversationInitialMessageUntilStreamConnect() {
 		if (!this.provider || this.provider.name !== "twilio" || !this.config.streaming.enabled) return false;
 		const streamAwareProvider = this.provider;
@@ -1913,8 +1945,7 @@ var CallManager = class {
 	async getCallFromMemoryOrStore(callId) {
 		const active = this.getCall(callId) ?? this.getCallByProviderCallId(callId);
 		if (active) return active;
-		const persisted = await findCallMatchesInStore(this.storePath, callId);
-		return persisted.byCallId ?? persisted.byProviderCallId;
+		return findCallInStore(this.storePath, callId);
 	}
 	/**
 	* Get call history (from persisted logs).
@@ -2106,7 +2137,7 @@ function formatBoundedChildOutput(output) {
 }
 //#endregion
 //#region extensions/voice-call/src/webhook/tailscale.ts
-const TAILSCALE_COMMAND_STDOUT_MAX_BYTES = 4 * 1024 * 1024;
+const TAILSCALE_COMMAND_STDOUT_MAX_BYTES = 4194304;
 const DEFAULT_TAILSCALE_HTTPS_PORT = 443;
 function buildTailscaleExposureArgs(opts) {
 	if (!opts.localUrl && opts.port === DEFAULT_TAILSCALE_HTTPS_PORT) return [
@@ -2240,7 +2271,6 @@ async function cleanupTailscaleExposure(config) {
 //#region extensions/voice-call/src/tunnel.ts
 const NGROK_LOG_BUFFER_MAX_CHARS = 16384;
 const NGROK_ERROR_MARKER = "ERR_NGROK";
-const TUNNEL_COMMAND_OUTPUT_MAX_BYTES = 16384;
 const NGROK_STOP_GRACE_MS = 2e3;
 const NGROK_FORCE_KILL_WAIT_MS = 1e3;
 async function terminateNgrokProcess(proc, isClosed) {
@@ -2280,11 +2310,6 @@ function listenForChildStreamErrors(proc, onError) {
 * // Later: await tunnel.stop();
 */
 async function startNgrokTunnel(config) {
-	if (config.authToken) await runNgrokCommand([
-		"config",
-		"add-authtoken",
-		config.authToken
-	]);
 	const args = [
 		"http",
 		String(config.port),
@@ -2295,11 +2320,17 @@ async function startNgrokTunnel(config) {
 	];
 	if (config.domain) args.push("--domain", config.domain);
 	return new Promise((resolve, reject) => {
-		const proc = spawn("ngrok", args, { stdio: [
-			"ignore",
-			"pipe",
-			"pipe"
-		] });
+		const proc = spawn("ngrok", args, {
+			stdio: [
+				"ignore",
+				"pipe",
+				"pipe"
+			],
+			...config.authToken ? { env: {
+				...process.env,
+				NGROK_AUTHTOKEN: config.authToken
+			} } : {}
+		});
 		let startupSettled = false;
 		let childClosed = false;
 		let publicUrl = null;
@@ -2372,27 +2403,6 @@ async function startNgrokTunnel(config) {
 	});
 }
 /**
-* Run an ngrok command and wait for completion.
-*/
-async function runNgrokCommand(args) {
-	const result = await runCommandWithTimeout(["ngrok", ...args], {
-		killProcessTree: true,
-		maxOutputBytes: TUNNEL_COMMAND_OUTPUT_MAX_BYTES,
-		outputCapture: "tail",
-		timeoutMs: 3e4
-	});
-	if (result.termination === "timeout") throw new Error("ngrok command timed out");
-	if (result.code === 0) return result.stdout;
-	const output = result.stderr ? {
-		text: result.stderr,
-		truncated: Boolean(result.stderrTruncatedBytes)
-	} : {
-		text: result.stdout,
-		truncated: Boolean(result.stdoutTruncatedBytes)
-	};
-	throw new Error(`ngrok command failed: ${formatBoundedChildOutput(output)}`);
-}
-/**
 * Start a Tailscale serve/funnel tunnel.
 */
 async function startTailscaleTunnel(config) {
@@ -2527,8 +2537,8 @@ const DEFAULT_PRE_START_TIMEOUT_MS = 5e3;
 const DEFAULT_MAX_PENDING_CONNECTIONS = 32;
 const DEFAULT_MAX_PENDING_CONNECTIONS_PER_IP = 4;
 const DEFAULT_MAX_CONNECTIONS = 128;
-const MAX_INBOUND_MESSAGE_BYTES = 64 * 1024;
-const MAX_WS_BUFFERED_BYTES = 1024 * 1024;
+const MAX_INBOUND_MESSAGE_BYTES = 65536;
+const MAX_WS_BUFFERED_BYTES = 1048576;
 const MAX_PENDING_TTS_OPERATIONS_PER_STREAM = 8;
 const MAX_IGNORED_PLAYBACK_MARKS_PER_STREAM = 64;
 const PLAYBACK_MARK_TIMEOUT_GRACE_MS = 2e3;
@@ -2691,10 +2701,7 @@ var MediaStreamHandler = class {
 							session = null;
 						}
 						break;
-					case "mark":
-						if (session && message.mark?.name) this.acknowledgePlaybackMark(session.streamSid, message.mark.name);
-						break;
-					case "clear": break;
+					case "mark": if (session && message.mark?.name) this.acknowledgePlaybackMark(session.streamSid, message.mark.name);
 				}
 			} catch (error) {
 				console.error("[MediaStream] Error processing message:", error);
@@ -2756,7 +2763,7 @@ var MediaStreamHandler = class {
 							role: "user"
 						}
 					});
-					this.config.onPartialTranscript?.(callSid, partial);
+					this.config.onPartialTranscript?.(callSid, partial, streamSid);
 				},
 				onTranscript: (transcript) => {
 					const session = this.sessions.get(streamSid);
@@ -2783,12 +2790,12 @@ var MediaStreamHandler = class {
 							}
 						});
 					}
-					this.config.onTranscript?.(callSid, transcript);
+					this.config.onTranscript?.(callSid, transcript, streamSid);
 				},
 				onSpeechStart: () => {
 					const session = this.sessions.get(streamSid);
 					if (session) this.ensureActiveTurn(session);
-					this.config.onSpeechStart?.(callSid);
+					this.config.onSpeechStart?.(callSid, streamSid);
 				},
 				onError: (error) => {
 					console.warn("[MediaStream] Transcription session error:", error.message);
@@ -3362,7 +3369,7 @@ const MAX_WEBHOOK_BODY_BYTES = WEBHOOK_BODY_READ_DEFAULTS.preAuth.maxBytes;
 const WEBHOOK_BODY_TIMEOUT_MS = WEBHOOK_BODY_READ_DEFAULTS.preAuth.timeoutMs;
 const MISSING_REMOTE_ADDRESS_IN_FLIGHT_KEY = "__voice_call_no_remote__";
 const loadRealtimeTranscriptionRuntime = createLazyRuntimeModule(() => import("./realtime-transcription.runtime-CbJAs5t_.js"));
-const loadResponseGeneratorModule = createLazyRuntimeModule(() => import("./response-generator-tjyX4tDq.js"));
+const loadResponseGeneratorModule = createLazyRuntimeModule(() => import("./response-generator-BPKgHnuI.js"));
 function appendRecentTalkEventMetadata(call, event) {
 	const metadata = call.metadata ?? {};
 	const recent = Array.isArray(metadata.recentTalkEvents) ? metadata.recentTalkEvents.filter((entry) => Boolean(entry) && typeof entry === "object" && !Array.isArray(entry)) : [];
@@ -3419,7 +3426,7 @@ function buildTwilioReplayTwiML() {
 		body: "<?xml version=\"1.0\" encoding=\"UTF-8\"?><Response></Response>"
 	};
 }
-const WEBHOOK_REPLAY_RESPONSE_TTL_MS = 600 * 1e3;
+const WEBHOOK_REPLAY_RESPONSE_TTL_MS = 6e5;
 const WEBHOOK_REPLAY_RESPONSE_MAX_ENTRIES = 1e4;
 const WEBHOOK_REPLAY_RESPONSE_PRUNE_INTERVAL = 64;
 function cloneWebhookResponsePayload(payload) {
@@ -3516,6 +3523,21 @@ var VoiceCallWebhookServer = class {
 		if ((call.metadata?.mode ?? "conversation") !== "conversation") return false;
 		return (normalizeOptionalString(call.metadata?.initialMessage) ?? "").length > 0;
 	}
+	getCurrentStream(providerCallId, streamSid) {
+		if (this.provider.name !== "twilio") return;
+		const provider = this.provider;
+		const call = this.manager.getCallByProviderCallId(providerCallId);
+		return call && provider.hasRegisteredStream(providerCallId, streamSid) ? {
+			call,
+			provider
+		} : void 0;
+	}
+	interruptStreamReply(providerCallId, streamSid) {
+		const current = this.getCurrentStream(providerCallId, streamSid);
+		if (!current || this.shouldSuppressBargeInForInitialMessage(current.call)) return;
+		this.manager.invalidateAutoResponse(current.call);
+		current.provider.clearTtsQueue(providerCallId);
+	}
 	/**
 	* Initialize media streaming with the selected realtime transcription provider.
 	*/
@@ -3529,7 +3551,7 @@ var VoiceCallWebhookServer = class {
 			cfg: pluginConfig,
 			cfgForResolve: pluginConfig ?? {},
 			getConfiguredProvider: (providerId) => getRealtimeTranscriptionProvider(providerId, pluginConfig),
-			listProviders: () => listRealtimeTranscriptionProviders(pluginConfig),
+			listProviders: () => listRealtimeTranscriptionProviders(pluginConfig, Object.keys(streaming.providers)),
 			resolveProviderConfig: ({ provider, cfg, rawConfig }) => provider.resolveConfig?.({
 				cfg,
 				rawConfig
@@ -3572,18 +3594,19 @@ var VoiceCallWebhookServer = class {
 				}
 				return true;
 			},
-			onTranscript: (providerCallId, transcript) => {
+			onTranscript: (providerCallId, transcript, streamSid) => {
 				this.logger.info(`Transcript received ${providerCallId} chars=${transcript.length}`);
-				const call = this.manager.getCallByProviderCallId(providerCallId);
-				if (!call) {
-					this.logger.warn(`No active call found for provider ID: ${providerCallId}`);
+				const current = this.getCurrentStream(providerCallId, streamSid);
+				if (!current) {
+					this.logger.info(`Ignoring transcript from inactive stream ${streamSid} (${providerCallId})`);
 					return;
 				}
+				const { call, provider: streamProvider } = current;
 				if (this.shouldSuppressBargeInForInitialMessage(call)) {
 					this.logger.info(`Ignoring barge transcript while initial message is still playing (${providerCallId})`);
 					return;
 				}
-				if (this.provider.name === "twilio") this.provider.clearTtsQueue(providerCallId);
+				streamProvider.clearTtsQueue(providerCallId);
 				const event = {
 					id: `stream-transcript-${Date.now()}`,
 					type: "call.speech",
@@ -3595,32 +3618,36 @@ var VoiceCallWebhookServer = class {
 				};
 				this.processEventWithAutoResponse(event);
 			},
-			onSpeechStart: (providerCallId) => {
-				if (this.provider.name !== "twilio") return;
-				const call = this.manager.getCallByProviderCallId(providerCallId);
-				if (this.shouldSuppressBargeInForInitialMessage(call)) return;
-				this.provider.clearTtsQueue(providerCallId);
-			},
-			onPartialTranscript: (callId, partial) => {
+			onSpeechStart: (providerCallId, streamSid) => this.interruptStreamReply(providerCallId, streamSid),
+			onPartialTranscript: (callId, partial, streamSid) => {
 				this.logger.info(`Partial transcript ${callId} chars=${partial.length}`);
+				this.interruptStreamReply(callId, streamSid);
 			},
-			onTalkEvent: (providerCallId, _streamSid, event) => {
-				const call = this.manager.getCallByProviderCallId(providerCallId);
-				if (call) appendRecentTalkEventMetadata(call, event);
+			onTalkEvent: (providerCallId, streamSid, event) => {
+				const current = this.getCurrentStream(providerCallId, streamSid);
+				if (current) appendRecentTalkEventMetadata(current.call, event);
 			},
 			onConnect: (callId, streamSid) => {
 				this.logger.info(`Media stream connected: ${callId} -> ${streamSid}`);
 				this.streamDisconnectGrace.connect(callId, streamSid);
+				const call = this.manager.getCallByProviderCallId(callId);
+				if (call) this.manager.invalidateAutoResponse(call);
 				if (this.provider.name === "twilio") this.provider.registerCallStream(callId, streamSid);
 			},
-			onTranscriptionReady: (callId) => {
+			onTranscriptionReady: (callId, streamSid) => {
+				if (!this.getCurrentStream(callId, streamSid)) return;
 				this.manager.speakInitialMessage(callId).catch((err) => {
 					this.logger.warn(`Failed to speak initial message: ${String(err)}`);
 				});
 			},
 			onDisconnect: (callId, streamSid) => {
 				this.logger.info(`Media stream disconnected: ${callId} (${streamSid})`);
-				if (this.provider.name === "twilio") this.provider.unregisterCallStream(callId, streamSid);
+				if (this.provider.name === "twilio") {
+					const twilio = this.provider;
+					twilio.unregisterCallStream(callId, streamSid);
+					const call = this.manager.getCallByProviderCallId(callId);
+					if (call && !twilio.hasRegisteredStream(callId)) this.manager.invalidateAutoResponse(call);
+				}
 				this.streamDisconnectGrace.disconnect(callId, streamSid);
 			}
 		};
@@ -4018,8 +4045,18 @@ var VoiceCallWebhookServer = class {
 			this.logger.warn("Agent runtime missing; skipping auto-response");
 			return;
 		}
+		const response = this.manager.createAutoResponseGuard(call);
+		const speakResponse = async (text) => {
+			if (!response.isCurrent()) {
+				this.logger.info(`Discarding superseded automatic reply ${callId}`);
+				return false;
+			}
+			this.logger.info(`AI response queued ${callId} chars=${text.length}`);
+			return (await this.manager.speak(callId, text, { listenAfterPlayback: true })).success;
+		};
 		try {
 			const { generateVoiceResponse } = await loadResponseGeneratorModule();
+			if (!response.isCurrent()) return;
 			const numberRouteKey = resolveVoiceCallNumberRouteKeyForCall(call);
 			const effectiveConfig = resolveVoiceCallEffectiveConfig(this.config, numberRouteKey).config;
 			const result = await generateVoiceResponse({
@@ -4033,21 +4070,17 @@ var VoiceCallWebhookServer = class {
 				agentId: resolveCallAgentId(call, effectiveConfig),
 				transcript: call.transcript,
 				userMessage,
-				onEarlyText: async (text) => {
-					this.logger.info(`Early AI response queued ${callId} chars=${text.length}`);
-					return (await this.manager.speak(callId, text, { listenAfterPlayback: true })).success;
-				}
+				onEarlyText: speakResponse
 			});
 			if (result.error) {
 				this.logger.error(`Response generation error: ${result.error}`);
 				return;
 			}
-			if (result.text && !result.deliveredEarly) {
-				this.logger.info(`AI response delivered ${callId} chars=${result.text.length}`);
-				await this.manager.speak(callId, result.text, { listenAfterPlayback: true });
-			}
+			if (result.text && !result.deliveredEarly) await speakResponse(result.text);
 		} catch (err) {
 			this.logger.error(`Auto-response error: ${String(err)}`);
+		} finally {
+			response.release();
 		}
 	}
 };
@@ -4061,12 +4094,12 @@ const REALTIME_VOICE_CONSULT_SYSTEM_PROMPT = [
 	"Do not print secret values or dump environment variables; only check whether required configuration is present.",
 	"Be accurate, brief, and speakable."
 ].join(" ");
-const loadTelnyxProvider = createLazyRuntimeModule(() => import("./telnyx-DO8dFQcb.js"));
-const loadTwilioProvider = createLazyRuntimeModule(() => import("./twilio-JDt_1nlz.js"));
-const loadPlivoProvider = createLazyRuntimeModule(() => import("./plivo-BbVnahnQ.js"));
-const loadMockProvider = createLazyRuntimeModule(() => import("./mock-Dv7KxYGt.js"));
+const loadTelnyxProvider = createLazyRuntimeModule(() => import("./telnyx-BWmhDBtd.js"));
+const loadTwilioProvider = createLazyRuntimeModule(() => import("./twilio-78mzH8Lk.js"));
+const loadPlivoProvider = createLazyRuntimeModule(() => import("./plivo-CgYWuWVr.js"));
+const loadMockProvider = createLazyRuntimeModule(() => import("./mock-C57qj18W.js"));
 const loadRealtimeVoiceRuntime = createLazyRuntimeModule(() => import("./realtime-voice.runtime-vtdCOWg-.js"));
-const loadRealtimeHandler = createLazyRuntimeModule(() => import("./realtime-handler-BaMEKbmJ.js"));
+const loadRealtimeHandler = createLazyRuntimeModule(() => import("./realtime-handler-oehCkuE2.js"));
 function resolveVoiceCallConsultSessionKey(call) {
 	return resolveVoiceCallSessionKey({
 		config: call.config,

package/dist/runtime-entry.js

@@ -1,2 +1,2 @@
-import { t as createVoiceCallRuntime } from "./runtime-entry-DK8kPCGC.js";
+import { t as createVoiceCallRuntime } from "./runtime-entry-CJkbbEgg.js";
 export { createVoiceCallRuntime };

package/dist/setup-api.js

@@ -1,19 +1,6 @@
 import { asFiniteNumber, asOptionalRecord, isRecord, readStringField } from "openclaw/plugin-sdk/string-coerce-runtime";
 import { definePluginEntry } from "openclaw/plugin-sdk/plugin-entry";
 //#region extensions/voice-call/src/config-migration.ts
-/** Merge legacy provider-specific values into the canonical providers map. */
-function mergeProviderConfig(providersValue, providerId, compatValues) {
-	if (Object.keys(compatValues).length === 0) return asOptionalRecord(providersValue);
-	const providers = asOptionalRecord(providersValue) ?? {};
-	const existing = asOptionalRecord(providers[providerId]) ?? {};
-	return {
-		...providers,
-		[providerId]: {
-			...existing,
-			...compatValues
-		}
-	};
-}
 /** Migrate legacy voice-call config input to the current canonical shape. */
 function migrateVoiceCallLegacyConfigInput(params) {
 	const raw = asOptionalRecord(params.value) ?? {};
@@ -22,28 +9,71 @@ function migrateVoiceCallLegacyConfigInput(params) {
 	const twilio = asOptionalRecord(raw.twilio);
 	const streaming = asOptionalRecord(raw.streaming);
 	const configPathPrefix = params.configPathPrefix ?? "plugins.entries.voice-call.config";
-	const legacyStreamingOpenAICompat = {};
-	const streamingOpenAIApiKey = readStringField(streaming, "openaiApiKey");
-	if (streamingOpenAIApiKey) legacyStreamingOpenAICompat.apiKey = streamingOpenAIApiKey;
-	const streamingSttModel = readStringField(streaming, "sttModel");
-	if (streamingSttModel) legacyStreamingOpenAICompat.model = streamingSttModel;
-	const streamingSilenceDurationMs = asFiniteNumber(streaming?.silenceDurationMs);
-	if (streamingSilenceDurationMs !== void 0) legacyStreamingOpenAICompat.silenceDurationMs = streamingSilenceDurationMs;
-	const streamingVadThreshold = asFiniteNumber(streaming?.vadThreshold);
-	if (streamingVadThreshold !== void 0) legacyStreamingOpenAICompat.vadThreshold = streamingVadThreshold;
+	const changes = [];
+	if (raw.provider === "log") changes.push(`Moved ${configPathPrefix}.provider "log" → "mock".`);
+	if (typeof twilio?.from === "string") {
+		const source = `${configPathPrefix}.twilio.from`;
+		const target = `${configPathPrefix}.fromNumber`;
+		changes.push(raw.fromNumber != null ? `Removed ${source} (kept ${target}).` : `Moved ${source} → ${target}.`);
+	}
 	const streamingProvider = readStringField(streaming, "provider");
 	const legacyStreamingProvider = readStringField(streaming, "sttProvider");
 	const normalizedStreaming = streaming ? {
 		...streaming,
-		provider: streamingProvider ?? legacyStreamingProvider,
-		providers: mergeProviderConfig(streaming.providers, "openai", legacyStreamingOpenAICompat)
+		provider: streamingProvider ?? legacyStreamingProvider
 	} : void 0;
 	if (normalizedStreaming) {
 		delete normalizedStreaming.sttProvider;
-		delete normalizedStreaming.openaiApiKey;
-		delete normalizedStreaming.sttModel;
-		delete normalizedStreaming.silenceDurationMs;
-		delete normalizedStreaming.vadThreshold;
+		if (legacyStreamingProvider !== void 0) {
+			const source = `${configPathPrefix}.streaming.sttProvider`;
+			const target = `${configPathPrefix}.streaming.provider`;
+			changes.push(streamingProvider !== void 0 ? `Removed ${source} (kept ${target}).` : `Moved ${source} → ${target}.`);
+		}
+		for (const [legacyKey, canonicalKey, value] of [
+			[
+				"openaiApiKey",
+				"apiKey",
+				readStringField(streaming, "openaiApiKey")
+			],
+			[
+				"sttModel",
+				"model",
+				readStringField(streaming, "sttModel")
+			],
+			[
+				"silenceDurationMs",
+				"silenceDurationMs",
+				asFiniteNumber(streaming?.silenceDurationMs)
+			],
+			[
+				"vadThreshold",
+				"vadThreshold",
+				asFiniteNumber(streaming?.vadThreshold)
+			]
+		]) {
+			if (!Object.hasOwn(normalizedStreaming, legacyKey)) continue;
+			delete normalizedStreaming[legacyKey];
+			const source = `${configPathPrefix}.streaming.${legacyKey}`;
+			if (value === void 0 || value === "") {
+				changes.push(`Removed invalid ${source}.`);
+				continue;
+			}
+			const providers = asOptionalRecord(normalizedStreaming.providers);
+			const existing = asOptionalRecord(providers?.openai);
+			const target = `${configPathPrefix}.streaming.providers.openai.${canonicalKey}`;
+			if (existing?.[canonicalKey] !== void 0) {
+				changes.push(`Removed ${source} (kept ${target}).`);
+				continue;
+			}
+			normalizedStreaming.providers = {
+				...providers,
+				openai: {
+					...existing,
+					[canonicalKey]: value
+				}
+			};
+			changes.push(`Moved ${source} → ${target}.`);
+		}
 	}
 	const normalizedTwilio = twilio ? { ...twilio } : void 0;
 	if (normalizedTwilio) delete normalizedTwilio.from;
@@ -61,16 +91,6 @@ function migrateVoiceCallLegacyConfigInput(params) {
 		streaming: normalizedStreaming,
 		realtime: normalizedRealtime
 	};
-	const changes = [];
-	if (raw.provider === "log") changes.push(`Moved ${configPathPrefix}.provider "log" → "mock".`);
-	if (typeof twilio?.from === "string" && typeof raw.fromNumber !== "string") changes.push(`Moved ${configPathPrefix}.twilio.from → ${configPathPrefix}.fromNumber.`);
-	if (typeof streaming?.sttProvider === "string") changes.push(`Moved ${configPathPrefix}.streaming.sttProvider → ${configPathPrefix}.streaming.provider.`);
-	if (typeof streaming?.openaiApiKey === "string") changes.push(`Moved ${configPathPrefix}.streaming.openaiApiKey → ${configPathPrefix}.streaming.providers.openai.apiKey.`);
-	if (typeof streaming?.sttModel === "string") changes.push(`Moved ${configPathPrefix}.streaming.sttModel → ${configPathPrefix}.streaming.providers.openai.model.`);
-	if (asFiniteNumber(streaming?.silenceDurationMs) !== void 0) changes.push(`Moved ${configPathPrefix}.streaming.silenceDurationMs → ${configPathPrefix}.streaming.providers.openai.silenceDurationMs.`);
-	else if (typeof streaming?.silenceDurationMs === "number") changes.push(`Removed invalid ${configPathPrefix}.streaming.silenceDurationMs.`);
-	if (asFiniteNumber(streaming?.vadThreshold) !== void 0) changes.push(`Moved ${configPathPrefix}.streaming.vadThreshold → ${configPathPrefix}.streaming.providers.openai.vadThreshold.`);
-	else if (typeof streaming?.vadThreshold === "number") changes.push(`Removed invalid ${configPathPrefix}.streaming.vadThreshold.`);
 	if (realtimeAgentContext && Object.hasOwn(realtimeAgentContext, "includeSystemPrompt")) changes.push(`Removed ${configPathPrefix}.realtime.agentContext.includeSystemPrompt.`);
 	return {
 		config,

package/dist/{store-path-51_oC_xn.js → store-path-C_z9zKnA.js}

@@ -133,14 +133,7 @@ const { setRuntime: setVoiceCallStateRuntime, tryGetRuntime: getOptionalVoiceCal
 //#region extensions/voice-call/src/manager/replay-keys.ts
 /** Match the existing voice-call webhook replay-cache cardinality. */
 const MAX_MANAGER_REPLAY_KEYS = 1e4;
-function pruneOldestSetEntries(keys, maxEntries) {
-	while (keys.size > maxEntries) {
-		const oldest = keys.values().next().value;
-		if (oldest === void 0) break;
-		keys.delete(oldest);
-	}
-}
-function pruneOldestMapEntries(keys, maxEntries) {
+function pruneOldestEntries(keys, maxEntries) {
 	while (keys.size > maxEntries) {
 		const oldest = keys.keys().next().value;
 		if (oldest === void 0) break;
@@ -150,7 +143,7 @@ function pruneOldestMapEntries(keys, maxEntries) {
 /** Remember one manager replay key without refreshing duplicate insertion order. */
 function rememberManagerReplayKey(keys, key, maxEntries = MAX_MANAGER_REPLAY_KEYS) {
 	keys.add(key);
-	pruneOldestSetEntries(keys, maxEntries);
+	pruneOldestEntries(keys, maxEntries);
 }
 /** Append one per-call replay key and retain only the newest bounded suffix. */
 function appendCallReplayKey(keys, key, maxEntries = 500) {
@@ -170,7 +163,7 @@ function reserveRejectedProviderCall(calls, providerCallId, maxEntries = MAX_MAN
 	if (calls.has(providerCallId)) return;
 	const reservation = Symbol(providerCallId);
 	calls.set(providerCallId, reservation);
-	pruneOldestMapEntries(calls, maxEntries);
+	pruneOldestEntries(calls, maxEntries);
 	return reservation;
 }
 /** Release a rejected-call reservation only when the failing attempt still owns it. */
@@ -189,7 +182,7 @@ const MAX_CALL_RECORD_EVENTS = 1e3;
 const CALL_RECORD_EVENT_META_MAX_ENTRIES = 1100;
 const CALL_RECORD_CHUNK_MAX_ENTRIES = 48048;
 /** Raw UTF-8 bytes stored per call record chunk before base64 encoding. */
-const RAW_CALL_RECORD_CHUNK_BYTES = 47 * 1024;
+const RAW_CALL_RECORD_CHUNK_BYTES = 48128;
 let callRecordEventSequence = 0;
 /** Return the pre-SQLite JSONL call log path for migration/compat checks. */
 function resolveVoiceCallLegacyCallLogPath(storePath) {
@@ -205,7 +198,7 @@ function resolvePluginStateEnv(storePath) {
 /** Open the plugin state stores when the runtime is available. */
 function createCallRecordStateStores(storePath) {
 	const runtime = getOptionalVoiceCallStateRuntime();
-	if (!runtime) return null;
+	if (!runtime) throw new Error("Voice Call state runtime not initialized");
 	const env = resolvePluginStateEnv(storePath);
 	return {
 		events: runtime.state.openSyncKeyedStore({
@@ -392,7 +385,6 @@ function readCallRecordEvents(stores) {
 function persistCallRecord(storePath, call) {
 	try {
 		const stores = createCallRecordStateStores(storePath);
-		if (!stores) throw new Error("Voice Call state runtime not initialized");
 		const order = nextCallRecordOrder();
 		registerCallRecordEvent(stores, buildNewEventKey(order), call, order);
 	} catch (err) {
@@ -444,13 +436,11 @@ function readCallHistoryFromStore(storePath) {
 	}
 	return [];
 }
-/** Find the newest retained snapshots matching each call identifier namespace. */
-async function findCallMatchesInStore(storePath, callId) {
-	const calls = readCallHistoryFromStore(storePath);
-	return {
-		byCallId: calls.findLast((call) => call.callId === callId),
-		byProviderCallId: calls.findLast((call) => call.providerCallId === callId)
-	};
+/** Resolve an internal ID or retained provider alias to its newest logical call snapshot. */
+function findCallInStore(storePath, callId) {
+	const calls = readCallRecordEvents(createCallRecordStateStores(storePath));
+	const match = calls.findLast((call) => call.callId === callId) ?? calls.findLast((call) => call.providerCallId === callId);
+	return match ? calls.findLast((call) => call.callId === match.callId) : void 0;
 }
 /** Return the newest persisted call history rows up to the requested limit. */
 async function getCallHistoryFromStore(storePath, limit = 50) {
@@ -464,4 +454,4 @@ function resolveDefaultVoiceCallStoreDir(env = process.env) {
 	return path.join(resolveStateDir(env), "voice-calls");
 }
 //#endregion
-export { releaseRejectedProviderCall as _, CALL_RECORD_EVENT_META_MAX_ENTRIES as a, setVoiceCallStateRuntime as b, buildVoiceCallLegacyJsonlEventKey as c, loadActiveCallsFromStore as d, parseVoiceCallRecordLine as f, appendCallReplayKey as g, resolveVoiceCallLegacyCallLogPath as h, CALL_RECORD_EVENT_CHUNKS_NAMESPACE as i, findCallMatchesInStore as l, prepareVoiceCallRecordForStorage as m, CALL_RECORD_CHUNK_MAX_ENTRIES as n, MAX_CALL_RECORD_EVENTS as o, persistCallRecord as p, CALL_RECORD_EVENTS_NAMESPACE as r, RAW_CALL_RECORD_CHUNK_BYTES as s, resolveDefaultVoiceCallStoreDir as t, getCallHistoryFromStore as u, rememberManagerReplayKey as v, TerminalStates as x, reserveRejectedProviderCall as y };
+export { releaseRejectedProviderCall as _, CALL_RECORD_EVENT_META_MAX_ENTRIES as a, setVoiceCallStateRuntime as b, buildVoiceCallLegacyJsonlEventKey as c, loadActiveCallsFromStore as d, parseVoiceCallRecordLine as f, appendCallReplayKey as g, resolveVoiceCallLegacyCallLogPath as h, CALL_RECORD_EVENT_CHUNKS_NAMESPACE as i, findCallInStore as l, prepareVoiceCallRecordForStorage as m, CALL_RECORD_CHUNK_MAX_ENTRIES as n, MAX_CALL_RECORD_EVENTS as o, persistCallRecord as p, CALL_RECORD_EVENTS_NAMESPACE as r, RAW_CALL_RECORD_CHUNK_BYTES as s, resolveDefaultVoiceCallStoreDir as t, getCallHistoryFromStore as u, rememberManagerReplayKey as v, TerminalStates as x, reserveRejectedProviderCall as y };