@ouro.bot/cli 0.1.0-alpha.560 → 0.1.0-alpha.562

package/README.md

@@ -105,7 +105,7 @@ Task docs do not live in this repo anymore. Planning and doing docs live in the
 - Human TTY commands share one CLI surface family: bare `ouro` opens the home deck, `ouro up` uses the boot checklist, `ouro connect`/`ouro auth verify`/`ouro repair` agree on provider and vault truth, and `ouro help`/`ouro whoami`/`ouro versions`/`ouro hatch` render through the same Ouro-branded wizard/guide language instead of raw transcript walls. Orientation commands such as root `ouro connect` may use shorter live probes, while startup and verification commands own durable readiness updates.
 - Human-facing CLI commands that can wait on browser auth, vault IO, daemon startup, daemon restart, provider checks, or connector setup use a shared progress checklist. If a cursor may blink for more than a few seconds, the command should print or animate the current step instead of going quiet.
 - CLI commands that mutate bundle config, such as vault setup or `ouro connect bluebubbles`, run bundle sync after the change when `sync.enabled` is true and report a compact `bundle sync:` line.
-- Voice is transcript-first: voice sessions use the ordinary `state/sessions/<friend>/voice/<key>.json` session path and appear in Ouro Mailbox as text transcripts. ElevenLabs API credentials live in portable `runtime/config` at `integrations.elevenLabsApiKey`; Whisper.cpp CLI/model paths live in the machine runtime item at `voice.whisperCliPath` and `voice.whisperModelPath`.
+- Voice is transcript-first: voice sessions use the ordinary `state/sessions/<friend>/voice/<key>.json` session path and appear in Ouro Mailbox as text transcripts. ElevenLabs API credentials live in portable `runtime/config` at `integrations.elevenLabsApiKey` and `integrations.elevenLabsVoiceId`; Whisper.cpp CLI/model paths live in the machine runtime item at `voice.whisperCliPath` and `voice.whisperModelPath`. Phone calls, browser meetings, and local microphone capture are transports under the single `voice` sense, not separate senses; the Twilio phone transport uses Twilio Record -> Whisper.cpp -> voice session -> ElevenLabs -> Twilio Play.
 - The daemon discovers bundles dynamically from `~/AgentBundles`.
 - `ouro status` reports version, last-updated time, discovered agents, senses, and workers.
 - `bundle-meta.json` tracks the runtime version that last touched a bundle.

package/changelog.json

@@ -1,6 +1,22 @@
 {
   "_note": "This changelog is maintained as part of the PR/version-bump workflow. Agent-curated, not auto-generated. Agents read this file directly via read_file to understand what changed between versions.",
   "versions": [
+    {
+      "version": "0.1.0-alpha.562",
+      "changes": [
+        "BlueBubbles outbound delivery now blocks narrow internal/meta markers such as `[surfaced from inner dialog]`, `[pending from ...]:`, routing-control prompt sections, and `<think>` tags before they can reach iMessage.",
+        "The guard fails closed across surface, proactive BlueBubbles sends, pending-drain retries, normal flush, and speak/flushNow paths so blocked internal text is logged instead of queued for later delivery.",
+        "Regression coverage now keeps ordinary user-facing prose about inner-dialog concepts deliverable while preventing the reported internal surfaced-thought leakage path."
+      ]
+    },
+    {
+      "version": "0.1.0-alpha.561",
+      "changes": [
+        "Voice golden-path orchestration now accepts meeting URLs, checks local BlackHole/Multi-Output readiness, runs Whisper.cpp STT, routes text through ordinary `voice` sessions, streams ElevenLabs TTS, and writes playback-ready audio artifacts.",
+        "Voice runtime edges now include default Node Whisper.cpp process execution, default Node WebSocket adaptation for ElevenLabs, and tested playback artifact handling while keeping credentials injected at runtime.",
+        "Shared sense turns now store sessions under the explicit agent's bundle instead of depending on process argv for the session path, and voice setup guidance now reflects meeting URL intake plus live-browser handoff limits truthfully."
+      ]
+    },
     {
       "version": "0.1.0-alpha.560",
       "changes": [

package/dist/heart/daemon/cli-exec.js

@@ -2540,6 +2540,10 @@ async function buildConnectMenu(agent, deps, onProgress) {
     const elevenLabsApiKey = runtimeConfig.ok
         ? readRuntimeConfigString(runtimeConfig.config, "integrations.elevenLabsApiKey")
         : null;
+    const elevenLabsVoiceId = runtimeConfig.ok
+        ? readRuntimeConfigString(runtimeConfig.config, "integrations.elevenLabsVoiceId")
+            ?? readRuntimeConfigString(runtimeConfig.config, "voice.elevenLabsVoiceId")
+        : null;
     const shouldVerifyPerplexity = runtimeConfig.ok && !!perplexityApiKey;
     const shouldVerifyEmbeddings = runtimeConfig.ok && !!embeddingsApiKey;
     let perplexityVerification;
@@ -2613,6 +2617,7 @@ async function buildConnectMenu(agent, deps, onProgress) {
     const voiceStatus = runtimeConfig.ok
         ? machineRuntime.ok
             ? elevenLabsApiKey
+                && elevenLabsVoiceId
                 && hasRuntimeConfigValue(machineRuntime.config, "voice.whisperCliPath")
                 && hasRuntimeConfigValue(machineRuntime.config, "voice.whisperModelPath")
                 && voiceEnabled
@@ -2719,6 +2724,7 @@ async function buildConnectMenu(agent, deps, onProgress) {
             detailLines: runtimeConfig.ok && machineRuntime.ok
                 ? [
                     elevenLabsApiKey ? "ElevenLabs API key saved in portable runtime config" : "missing integrations.elevenLabsApiKey",
+                    elevenLabsVoiceId ? "ElevenLabs voice ID saved in portable runtime config" : "missing integrations.elevenLabsVoiceId",
                     hasRuntimeConfigValue(machineRuntime.config, "voice.whisperCliPath") ? "Whisper.cpp CLI path saved for this machine" : "missing voice.whisperCliPath",
                     hasRuntimeConfigValue(machineRuntime.config, "voice.whisperModelPath") ? "Whisper.cpp model path saved for this machine" : "missing voice.whisperModelPath",
                 ]
@@ -4270,13 +4276,22 @@ function connectMenuTarget(answer) {
 async function executeConnectVoice(agent, deps) {
     const message = [
         `Voice foundation for ${agent}`,
-        "Configure the portable ElevenLabs API key with:",
+        "Configure portable ElevenLabs settings with:",
         `  ouro vault config set --agent ${agent} --key integrations.elevenLabsApiKey`,
+        `  ouro vault config set --agent ${agent} --key integrations.elevenLabsVoiceId`,
         "Configure this machine's Whisper.cpp attachment with:",
         `  ouro vault config set --agent ${agent} --scope machine --key voice.whisperCliPath`,
         `  ouro vault config set --agent ${agent} --scope machine --key voice.whisperModelPath`,
+        "Optional Twilio phone testing setup:",
+        `  ouro vault config set --agent ${agent} --key voice.twilioAccountSid`,
+        `  ouro vault config set --agent ${agent} --key voice.twilioAuthToken`,
+        `  ouro vault config set --agent ${agent} --scope machine --key voice.twilioPublicUrl`,
+        `  ouro vault config set --agent ${agent} --scope machine --key voice.twilioPort --value 18910`,
+        `  ouro vault config set --agent ${agent} --scope machine --key voice.twilioDefaultFriendId --value ari`,
+        `  node dist/senses/voice-twilio-entry.js --agent ${agent} --port 18910 --public-url https://<cloudflare-tunnel>`,
+        `Set the Twilio number's Voice webhook to POST https://<cloudflare-tunnel>/voice/twilio/incoming.`,
         "Then enable agent.json: senses.voice.enabled = true and restart with `ouro up`.",
-        "Meeting-link joining and browser/system audio routing are tracked as the next milestone.",
+        "Meeting links use URL intake plus BlackHole/Multi-Output readiness checks. Phone testing uses Twilio Record -> Whisper.cpp -> voice session -> ElevenLabs -> Twilio Play.",
     ].join("\n");
     deps.writeStdout(message);
     return message;

package/dist/heart/daemon/sense-manager.js

@@ -197,6 +197,9 @@ function senseFactsFromRuntimeConfig(agent, senses, runtimeConfig, machineRuntim
         const missing = [];
         if (!textField(integrations, "elevenLabsApiKey"))
             missing.push("integrations.elevenLabsApiKey");
+        if (!textField(integrations, "elevenLabsVoiceId") && !textField(payload.voice, "elevenLabsVoiceId")) {
+            missing.push("integrations.elevenLabsVoiceId");
+        }
         if (!textField(voice, "whisperCliPath"))
             missing.push("voice.whisperCliPath");
         if (!textField(voice, "whisperModelPath"))

package/dist/heart/turn-context.js

@@ -165,6 +165,7 @@ function readSenseStatusLines() {
     const bluebubbles = recordOrUndefined(machinePayload.bluebubbles) ?? recordOrUndefined(payload.bluebubbles);
     const mailroom = recordOrUndefined(runtimePayload.mailroom) ?? recordOrUndefined(payload.mailroom);
     const voice = recordOrUndefined(machinePayload.voice) ?? recordOrUndefined(payload.voice);
+    const portableVoice = recordOrUndefined(runtimePayload.voice) ?? recordOrUndefined(payload.voice);
     const integrations = recordOrUndefined(runtimePayload.integrations) ?? recordOrUndefined(payload.integrations);
     const privateKeys = mailroom?.privateKeys;
     const configured = {
@@ -172,7 +173,10 @@ function readSenseStatusLines() {
         teams: hasTextField(teams, "clientId") && hasTextField(teams, "clientSecret") && hasTextField(teams, "tenantId"),
         bluebubbles: hasTextField(bluebubbles, "serverUrl") && hasTextField(bluebubbles, "password"),
         mail: hasTextField(mailroom, "mailboxAddress") && !!privateKeys && typeof privateKeys === "object" && !Array.isArray(privateKeys),
-        voice: hasTextField(integrations, "elevenLabsApiKey") && hasTextField(voice, "whisperCliPath") && hasTextField(voice, "whisperModelPath"),
+        voice: hasTextField(integrations, "elevenLabsApiKey")
+            && (hasTextField(integrations, "elevenLabsVoiceId") || hasTextField(portableVoice, "elevenLabsVoiceId"))
+            && hasTextField(voice, "whisperCliPath")
+            && hasTextField(voice, "whisperModelPath"),
     };
     const rows = [
         { label: "CLI", status: "interactive" },

package/dist/mind/prompt.js

@@ -441,6 +441,7 @@ function localSenseStatusLines() {
     const bluebubbles = recordOrUndefined(machinePayload.bluebubbles) ?? recordOrUndefined(payload.bluebubbles);
     const mailroom = recordOrUndefined(runtimePayload.mailroom) ?? recordOrUndefined(payload.mailroom);
     const voice = recordOrUndefined(machinePayload.voice) ?? recordOrUndefined(payload.voice);
+    const portableVoice = recordOrUndefined(runtimePayload.voice) ?? recordOrUndefined(payload.voice);
     const integrations = recordOrUndefined(runtimePayload.integrations) ?? recordOrUndefined(payload.integrations);
     const privateKeys = mailroom?.privateKeys;
     const configured = {
@@ -448,7 +449,10 @@ function localSenseStatusLines() {
         teams: hasTextField(teams, "clientId") && hasTextField(teams, "clientSecret") && hasTextField(teams, "tenantId"),
         bluebubbles: hasTextField(bluebubbles, "serverUrl") && hasTextField(bluebubbles, "password"),
         mail: hasTextField(mailroom, "mailboxAddress") && !!privateKeys && typeof privateKeys === "object" && !Array.isArray(privateKeys),
-        voice: hasTextField(integrations, "elevenLabsApiKey") && hasTextField(voice, "whisperCliPath") && hasTextField(voice, "whisperModelPath"),
+        voice: hasTextField(integrations, "elevenLabsApiKey")
+            && (hasTextField(integrations, "elevenLabsVoiceId") || hasTextField(portableVoice, "elevenLabsVoiceId"))
+            && hasTextField(voice, "whisperCliPath")
+            && hasTextField(voice, "whisperModelPath"),
     };
     const rows = [
         { label: "CLI", status: "interactive" },
@@ -502,7 +506,7 @@ function senseRuntimeGuidance(channel, preReadStatusLines) {
     lines.push("mail validation diagnostics: health checks, bounded mail tools, access logs, and UI inspection can support validation, but they are evidence inside those paths, not additional paths. If asked to name golden paths, do not include diagnostic commands, tool names, or status checks in the answer.");
     lines.push("mail diagnostic naming: `ouro doctor` is installation-wide; do not invent `ouro doctor --agent <agent>`.");
     lines.push("mail setup boundaries: do not invent `ouro auth verify --provider mail`, HEY OAuth, HEY IMAP, `ouro mcp call mail ...`, policy flags, autonomous sending, destructive mail actions, or production MX/DNS/forwarding changes. HEY export, HEY forwarding, DNS, MX cutover, sending, and destructive actions require explicit human confirmation.");
-    lines.push("voice setup truth: voice sessions are transcript-first local sessions. ElevenLabs credentials belong in portable runtime/config at `integrations.elevenLabsApiKey`; Whisper.cpp CLI/model paths belong in the machine runtime item under `voice.whisperCliPath` and `voice.whisperModelPath`. Meeting-link joining and browser/system audio routing are a later milestone, not current setup truth.");
+    lines.push("voice setup truth: voice sessions are transcript-first local sessions. ElevenLabs credentials belong in portable runtime/config at `integrations.elevenLabsApiKey` and `integrations.elevenLabsVoiceId`; Whisper.cpp CLI/model paths belong in the machine runtime item under `voice.whisperCliPath` and `voice.whisperModelPath`. Meeting links have URL intake and local BlackHole/Multi-Output readiness checks; phone testing uses Twilio Record -> Whisper.cpp -> voice session -> ElevenLabs -> Twilio Play. Live browser join/injection remains an explicit handoff edge until provider automation lands.");
     if (channel === "cli") {
         lines.push("cli is interactive: it is available when the user opens it, not something `ouro up` daemonizes.");
     }

package/dist/repertoire/tools-session.js

@@ -656,6 +656,12 @@ exports.sessionToolDefinitions = [
                                 detail: "bluebubbles could not resolve a routable target for that session",
                             };
                         }
+                        if (result.reason === "blocked_meta_content") {
+                            return {
+                                status: "blocked",
+                                detail: "blocked: contains internal meta markers",
+                            };
+                        }
                         if (result.reason === "send_error") {
                             return {
                                 status: "failed",

package/dist/repertoire/tools-surface.js

@@ -39,6 +39,8 @@ const identity_1 = require("../heart/identity");
 const surface_tool_1 = require("../senses/surface-tool");
 const obligations_1 = require("../arc/obligations");
 const session_activity_1 = require("../heart/session-activity");
+const bluebubbles_meta_guard_1 = require("../senses/bluebubbles-meta-guard");
+const runtime_1 = require("../nerves/runtime");
 const path = __importStar(require("path"));
 // Surface tool schema — canonical home. Handler lives in senses/surface-tool.ts.
 exports.surfaceToolDef = {
@@ -71,6 +73,21 @@ exports.surfaceToolDef = {
 exports.surfaceToolDefinition = {
     tool: exports.surfaceToolDef,
     handler: async (args, ctx) => {
+        const rawContent = args.content ?? "";
+        if ((0, bluebubbles_meta_guard_1.containsInternalMetaMarkers)(rawContent)) {
+            (0, runtime_1.emitNervesEvent)({
+                level: "warn",
+                component: "repertoire",
+                event: "tools.surface_meta_blocked",
+                message: "surface tool blocked: internal meta markers in content",
+                meta: {
+                    hasDelegationId: Boolean(args.delegationId),
+                    hasFriendId: Boolean(args.friendId),
+                    contentLength: rawContent.length,
+                },
+            });
+            return "failed — blocked: contains internal meta markers";
+        }
         const queue = ctx?.delegatedOrigins ?? [];
         const agentName = (() => { try {
             return (0, identity_1.getAgentName)();

package/dist/senses/bluebubbles/index.js

@@ -69,6 +69,7 @@ const prompt_1 = require("../../mind/prompt");
 const mcp_manager_1 = require("../../repertoire/mcp-manager");
 const runtime_1 = require("../../nerves/runtime");
 const proactive_content_guard_1 = require("../proactive-content-guard");
+const bluebubbles_meta_guard_1 = require("../bluebubbles-meta-guard");
 const model_1 = require("./model");
 const client_1 = require("./client");
 const inbound_log_1 = require("./inbound-log");
@@ -616,6 +617,17 @@ function createBlueBubblesCallbacks(client, chat, replyTarget, isGroupChat, onVi
             if (!trimmed)
                 return;
             textBuffer = "";
+            if ((0, bluebubbles_meta_guard_1.containsInternalMetaMarkers)(trimmed)) {
+                (0, bluebubbles_meta_guard_1.emitBluebubblesMetaBlocked)({
+                    site: "flushNow",
+                    message: "bluebubbles speak text blocked: internal meta markers",
+                    meta: {
+                        chatGuid: chat.chatGuid ?? null,
+                        messageLength: trimmed.length,
+                    },
+                });
+                return;
+            }
             await client.sendText({
                 chat,
                 text: trimmed,
@@ -650,6 +662,17 @@ function createBlueBubblesCallbacks(client, chat, replyTarget, isGroupChat, onVi
                 enqueue("typing_stop", async () => { await client.setTyping(chat, false); });
                 await queue;
             }
+            if ((0, bluebubbles_meta_guard_1.containsInternalMetaMarkers)(trimmed)) {
+                (0, bluebubbles_meta_guard_1.emitBluebubblesMetaBlocked)({
+                    site: "flush",
+                    message: "bluebubbles outbound text blocked: internal meta markers",
+                    meta: {
+                        chatGuid: chat.chatGuid ?? null,
+                        messageLength: trimmed.length,
+                    },
+                });
+                return;
+            }
             await client.sendText({
                 chat,
                 text: trimmed,
@@ -1925,6 +1948,17 @@ function buildChatRefForSessionKey(friend, sessionKey) {
     };
 }
 async function sendProactiveBlueBubblesMessageToSession(params, deps = {}) {
+    if ((0, bluebubbles_meta_guard_1.containsInternalMetaMarkers)(params.text)) {
+        (0, bluebubbles_meta_guard_1.emitBluebubblesMetaBlocked)({
+            site: "proactive",
+            message: "bluebubbles proactive send blocked: internal meta markers",
+            meta: {
+                friendId: params.friendId,
+                sessionKey: params.sessionKey,
+            },
+        });
+        return { delivered: false, reason: "blocked_meta_content" };
+    }
     const resolvedDeps = { ...defaultDeps, ...deps };
     const client = resolvedDeps.createClient();
     const store = resolvedDeps.createFriendStore();
@@ -2131,6 +2165,22 @@ async function drainAndSendPendingBlueBubbles(deps = {}, pendingRoot) {
             catch { /* ignore */ }
             continue;
         }
+        if ((0, bluebubbles_meta_guard_1.containsInternalMetaMarkers)(messageText)) {
+            result.skipped++;
+            try {
+                fs.unlinkSync(filePath);
+            }
+            catch { /* ignore */ }
+            (0, bluebubbles_meta_guard_1.emitBluebubblesMetaBlocked)({
+                site: "drain",
+                message: "bluebubbles drain blocked: internal meta markers",
+                meta: {
+                    friendId,
+                    filePath,
+                },
+            });
+            continue;
+        }
         const internalBlockReason = (0, proactive_content_guard_1.getProactiveInternalContentBlockReason)(messageText);
         if (internalBlockReason) {
             result.skipped++;

package/dist/senses/bluebubbles-meta-guard.js

@@ -0,0 +1,40 @@
+"use strict";
+// Outbound BlueBubbles meta-content guard.
+//
+// Blocks accidental delivery of internal/meta text — pipeline section markers,
+// surfacing-mechanics prefixes, reasoning tags — to the live iMessage channel.
+// Failure mode is "drop and log", never queue for later delivery.
+//
+// Patterns are deliberately narrow: bracketed system markers and angle-bracket
+// reasoning tags. Plain prose mentioning "inner dialog" or "attention queue"
+// is NOT blocked, so user-facing replies that legitimately discuss those
+// concepts still pass.
+Object.defineProperty(exports, "__esModule", { value: true });
+exports.containsInternalMetaMarkers = containsInternalMetaMarkers;
+exports.emitBluebubblesMetaBlocked = emitBluebubblesMetaBlocked;
+const runtime_1 = require("../nerves/runtime");
+const META_CONTENT_PATTERNS = [
+    /\[surfaced from inner dialog\]/i,
+    /\[pending from [^\]]+\]:/i,
+    /\[conversation scope:/i,
+    /\[recent active lanes\]/i,
+    /\[routing control:/i,
+    /<\/?think>/i,
+];
+function containsInternalMetaMarkers(text) {
+    if (!text)
+        return false;
+    return META_CONTENT_PATTERNS.some((pattern) => pattern.test(text));
+}
+function emitBluebubblesMetaBlocked(options) {
+    (0, runtime_1.emitNervesEvent)({
+        level: "warn",
+        component: "senses",
+        event: "senses.bluebubbles_meta_blocked",
+        message: options.message,
+        meta: {
+            site: options.site,
+            ...options.meta,
+        },
+    });
+}

package/dist/senses/shared-turn.js

@@ -43,6 +43,7 @@ exports.stripThinkBlocks = stripThinkBlocks;
 exports.runSenseTurn = runSenseTurn;
 const os = __importStar(require("os"));
 const path = __importStar(require("path"));
+const fs = __importStar(require("fs"));
 const core_1 = require("../heart/core");
 const identity_1 = require("../heart/identity");
 const config_1 = require("../heart/config");
@@ -130,7 +131,9 @@ async function runSenseTurn(options) {
     // Initialize MCP manager so MCP tools appear as first-class tools in the agent's tool list
     const mcpManager = await (0, mcp_manager_1.getSharedMcpManager)() ?? undefined;
     // Session path and loading
-    const sessPath = (0, config_1.sessionPath)(friendId, channel, sessionKey);
+    const sessionDir = path.join(agentRoot, "state", "sessions", friendId, channel);
+    fs.mkdirSync(sessionDir, { recursive: true });
+    const sessPath = path.join(sessionDir, `${(0, config_1.sanitizeKey)(sessionKey)}.json`);
     const existing = (0, context_1.loadSession)(sessPath);
     let sessionState = existing?.state;
     let persistPromise;

package/dist/senses/voice/audio-routing.js

@@ -0,0 +1,119 @@
+"use strict";
+Object.defineProperty(exports, "__esModule", { value: true });
+exports.createNodeVoiceCommandRunner = createNodeVoiceCommandRunner;
+exports.inspectVoiceAudioRouting = inspectVoiceAudioRouting;
+const child_process_1 = require("child_process");
+const runtime_1 = require("../../nerves/runtime");
+function createNodeVoiceCommandRunner() {
+    return (command, args, options) => new Promise((resolve, reject) => {
+        const child = (0, child_process_1.spawn)(command, args, { stdio: ["ignore", "pipe", "pipe"] });
+        const stdout = [];
+        const stderr = [];
+        const timer = setTimeout(() => {
+            child.kill("SIGTERM");
+            reject(new Error(`command timed out after ${options.timeoutMs}ms`));
+        }, options.timeoutMs);
+        child.stdout.on("data", (chunk) => stdout.push(chunk));
+        child.stderr.on("data", (chunk) => stderr.push(chunk));
+        child.on("error", (error) => {
+            clearTimeout(timer);
+            reject(error);
+        });
+        child.on("close", (exitCode) => {
+            clearTimeout(timer);
+            resolve({
+                stdout: Buffer.concat(stdout).toString("utf8"),
+                stderr: Buffer.concat(stderr).toString("utf8"),
+                exitCode: exitCode ?? 0,
+            });
+        });
+    });
+}
+function parseDeviceLines(output) {
+    return output
+        .split(/\r?\n/)
+        .map((line) => line.trim())
+        .filter(Boolean);
+}
+function commandFailureMessage(exitCode, result) {
+    const stderr = result.stderr?.trim();
+    if (stderr)
+        return stderr;
+    const stdout = result.stdout?.trim();
+    if (stdout)
+        return stdout;
+    return `exit ${exitCode}`;
+}
+function setupGuidance(missing, currentOutput, outputDeviceName) {
+    const guidance = missing.map((device) => `Install or configure the local audio device: ${device}.`);
+    if (currentOutput && currentOutput !== outputDeviceName) {
+        guidance.push(`Browser meeting audio should be routed through ${outputDeviceName}; current output is ${currentOutput}.`);
+    }
+    return guidance;
+}
+async function inspectVoiceAudioRouting(options = {}) {
+    const commandRunner = options.commandRunner ?? createNodeVoiceCommandRunner();
+    const switchAudioSourcePath = options.switchAudioSourcePath ?? "SwitchAudioSource";
+    const captureDeviceName = options.captureDeviceName ?? "BlackHole 2ch";
+    const outputDeviceName = options.outputDeviceName ?? "Multi-Output Device";
+    const timeoutMs = options.timeoutMs ?? 5_000;
+    try {
+        const devicesResult = await commandRunner(switchAudioSourcePath, ["-a"], { timeoutMs });
+        if (typeof devicesResult.exitCode === "number" && devicesResult.exitCode !== 0) {
+            throw new Error(commandFailureMessage(devicesResult.exitCode, devicesResult));
+        }
+        const currentResult = await commandRunner(switchAudioSourcePath, ["-c"], { timeoutMs });
+        if (typeof currentResult.exitCode === "number" && currentResult.exitCode !== 0) {
+            throw new Error(commandFailureMessage(currentResult.exitCode, currentResult));
+        }
+        const devices = parseDeviceLines(devicesResult.stdout ?? "");
+        const currentOutput = parseDeviceLines(currentResult.stdout ?? "")[0] ?? null;
+        const hasCaptureDevice = devices.includes(captureDeviceName);
+        const hasOutputDevice = devices.includes(outputDeviceName);
+        const missing = [
+            ...(hasCaptureDevice ? [] : [captureDeviceName]),
+            ...(hasOutputDevice ? [] : [outputDeviceName]),
+        ];
+        const result = {
+            status: missing.length === 0 ? "ready" : "needs_setup",
+            hasCaptureDevice,
+            hasOutputDevice,
+            currentOutput,
+            missing,
+            guidance: setupGuidance(missing, currentOutput, outputDeviceName),
+        };
+        (0, runtime_1.emitNervesEvent)({
+            component: "senses",
+            event: "senses.voice_audio_routing_checked",
+            message: "voice audio routing readiness checked",
+            meta: {
+                status: result.status,
+                hasCaptureDevice,
+                hasOutputDevice,
+                currentOutput,
+                missing,
+            },
+        });
+        return result;
+    }
+    catch (error) {
+        const message = error instanceof Error ? error.message : String(error);
+        const result = {
+            status: "unknown",
+            hasCaptureDevice: false,
+            hasOutputDevice: false,
+            currentOutput: null,
+            missing: [captureDeviceName, outputDeviceName],
+            guidance: setupGuidance([captureDeviceName, outputDeviceName], null, outputDeviceName),
+            error: message,
+        };
+        (0, runtime_1.emitNervesEvent)({
+            level: "error",
+            component: "senses",
+            event: "senses.voice_audio_routing_error",
+            message: "voice audio routing readiness check failed",
+            meta: { error: message, missing: result.missing },
+        });
+        return result;
+    }
+}

package/dist/senses/voice/elevenlabs.js

@@ -1,6 +1,7 @@
 "use strict";
 Object.defineProperty(exports, "__esModule", { value: true });
 exports.DEFAULT_ELEVENLABS_MIME_TYPE = exports.DEFAULT_ELEVENLABS_OUTPUT_FORMAT = exports.DEFAULT_ELEVENLABS_MODEL_ID = void 0;
+exports.createNodeElevenLabsSocketFactory = createNodeElevenLabsSocketFactory;
 exports.createElevenLabsTtsClient = createElevenLabsTtsClient;
 const runtime_1 = require("../../nerves/runtime");
 exports.DEFAULT_ELEVENLABS_MODEL_ID = "eleven_flash_v2_5";
@@ -14,15 +15,67 @@ function elevenLabsStreamUrl(voiceId, modelId, outputFormat) {
     return `wss://api.elevenlabs.io/v1/text-to-speech/${encodeURIComponent(voiceId)}/stream-input?${params.toString()}`;
 }
 function payloadText(payload) {
+    if (payload && typeof payload === "object" && "data" in payload) {
+        return payloadText(payload.data);
+    }
     if (typeof payload === "string")
         return payload;
     if (Buffer.isBuffer(payload))
         return payload.toString("utf8");
     return String(payload ?? "");
 }
+function createNodeElevenLabsSocketFactory(webSocketConstructor) {
+    const WebSocketConstructor = webSocketConstructor
+        ?? globalThis.WebSocket;
+    if (!WebSocketConstructor) {
+        throw new Error("global WebSocket is unavailable; inject an ElevenLabs socketFactory");
+    }
+    return (url) => {
+        const socket = new WebSocketConstructor(url);
+        const handlers = {
+            open: [],
+            message: [],
+            error: [],
+            close: [],
+        };
+        const emit = (event, payload) => {
+            for (const handler of handlers[event]) {
+                handler(payload);
+            }
+        };
+        const attach = (event) => {
+            const listener = (payload) => emit(event, payload);
+            if (typeof socket.addEventListener === "function") {
+                socket.addEventListener(event, listener);
+            }
+            else if (typeof socket.on === "function") {
+                socket.on(event, listener);
+            }
+            else {
+                socket[`on${event}`] = listener;
+            }
+        };
+        attach("open");
+        attach("message");
+        attach("error");
+        attach("close");
+        return {
+            on(event, handler) {
+                handlers[event].push(handler);
+            },
+            send(payload) {
+                socket.send(payload);
+            },
+            close() {
+                socket.close();
+            },
+        };
+    };
+}
 function createElevenLabsTtsClient(options) {
     const modelId = options.modelId ?? exports.DEFAULT_ELEVENLABS_MODEL_ID;
     const outputFormat = options.outputFormat ?? exports.DEFAULT_ELEVENLABS_OUTPUT_FORMAT;
+    const socketFactory = options.socketFactory ?? createNodeElevenLabsSocketFactory();
     const mimeType = outputFormat === exports.DEFAULT_ELEVENLABS_OUTPUT_FORMAT
         ? exports.DEFAULT_ELEVENLABS_MIME_TYPE
         : "audio/mpeg";
@@ -40,7 +93,7 @@ function createElevenLabsTtsClient(options) {
                 throw new Error("voice TTS text is empty");
             }
             const url = elevenLabsStreamUrl(options.voiceId, modelId, outputFormat);
-            const socket = options.socketFactory(url);
+            const socket = socketFactory(url);
             const chunks = [];
             (0, runtime_1.emitNervesEvent)({
                 component: "senses",