careervivid 2.1.13 → 2.1.16

package/dist/commands/agent/repl.d.ts.map

@@ -1 +1 @@
-{"version":3,"file":"repl.d.ts","sourceRoot":"","sources":["../../../src/commands/agent/repl.ts"],"names":[],"mappings":"AAIA,OAAO,EAAE,sBAAsB,EAAE,MAAM,uCAAuC,CAAC;AAC/E,OAAO,EAAE,WAAW,EAAE,MAAM,4BAA4B,CAAC;AAEzD,OAAO,EAA4D,KAAK,WAAW,EAAE,MAAM,iBAAiB,CAAC;AAO7G,wBAAgB,iBAAiB,CAAC,SAAS,EAAE,MAAM,GAAG,IAAI,EAAE,KAAK,GAAE,MAAM,GAAG,IAAW,QAwBtF;AAED,wBAAsB,OAAO,CAC3B,MAAM,EAAE,WAAW,GAAG,sBAAsB,GAAG,IAAI,EACnD,OAAO,EAAE;IAAE,OAAO,CAAC,EAAE,OAAO,CAAC;IAAC,KAAK,CAAC,EAAE,MAAM,CAAC;IAAC,UAAU,CAAC,EAAE,MAAM,CAAC;IAAC,OAAO,CAAC,EAAE,MAAM,CAAC;IAAC,SAAS,CAAC,EAAE,MAAM,CAAC;IAAC,MAAM,CAAC,EAAE,MAAM,CAAC;IAAC,IAAI,CAAC,EAAE,OAAO,CAAC;IAAC,MAAM,CAAC,EAAE,OAAO,CAAC;IAAC,MAAM,CAAC,EAAE,OAAO,CAAA;CAAE,EAC9K,gBAAgB,EAAE,WAAW,EAC7B,aAAa,EAAE,MAAM,EACrB,QAAQ,EAAE,MAAM,GAAG,SAAS,EAC5B,iBAAiB,EAAE,MAAM,EACzB,KAAK,EAAE,GAAG,EAAE,GACX,OAAO,CAAC,IAAI,CAAC,CAosBf"}
+{"version":3,"file":"repl.d.ts","sourceRoot":"","sources":["../../../src/commands/agent/repl.ts"],"names":[],"mappings":"AAIA,OAAO,EAAE,sBAAsB,EAAE,MAAM,uCAAuC,CAAC;AAC/E,OAAO,EAAE,WAAW,EAAE,MAAM,4BAA4B,CAAC;AAEzD,OAAO,EAA4D,KAAK,WAAW,EAAE,MAAM,iBAAiB,CAAC;AAS7G,wBAAgB,iBAAiB,CAAC,SAAS,EAAE,MAAM,GAAG,IAAI,EAAE,KAAK,GAAE,MAAM,GAAG,IAAW,QAwBtF;AAED,wBAAsB,OAAO,CAC3B,MAAM,EAAE,WAAW,GAAG,sBAAsB,GAAG,IAAI,EACnD,OAAO,EAAE;IAAE,OAAO,CAAC,EAAE,OAAO,CAAC;IAAC,KAAK,CAAC,EAAE,MAAM,CAAC;IAAC,UAAU,CAAC,EAAE,MAAM,CAAC;IAAC,OAAO,CAAC,EAAE,MAAM,CAAC;IAAC,SAAS,CAAC,EAAE,MAAM,CAAC;IAAC,MAAM,CAAC,EAAE,MAAM,CAAC;IAAC,IAAI,CAAC,EAAE,OAAO,CAAC;IAAC,MAAM,CAAC,EAAE,OAAO,CAAC;IAAC,MAAM,CAAC,EAAE,OAAO,CAAA;CAAE,EAC9K,gBAAgB,EAAE,WAAW,EAC7B,aAAa,EAAE,MAAM,EACrB,QAAQ,EAAE,MAAM,GAAG,SAAS,EAC5B,iBAAiB,EAAE,MAAM,EACzB,KAAK,EAAE,GAAG,EAAE,GACX,OAAO,CAAC,IAAI,CAAC,CA6uBf"}

package/dist/commands/agent/repl.js

@@ -4,10 +4,10 @@ import ora from "ora";
 import { isSafeCommand } from "../../agent/tools/coding.js";
 import { CareerVividProxyEngine } from "../../agent/CareerVividProxyEngine.js";
 import { CV_MODELS } from "./configurator.js";
-import { loadConfig, getGeminiKey, getProviderKey, setProviderKey } from "../../config.js";
+import { loadConfig, getProviderKey, setProviderKey } from "../../config.js";
 import { auditLog, writeSessionSummary, SESSION_ID } from "../../agent/agentAuditLog.js";
 import { runShellEscape } from "../../lib/shell.js";
-import { isVoiceEnabled, setVoiceEnabled, setLastResponse, getLastResponse, speakText, stopPlayback } from "../../lib/tts.js";
+import { isVoiceEnabled, setVoiceEnabled, setLastResponse, getLastResponse, speakText, stopPlayback, getCurrentVoice, setCurrentVoice, getCurrentTtsModel, setCurrentTtsModel, AVAILABLE_VOICES, AVAILABLE_TTS_MODELS } from "../../lib/tts.js";
 const { prompt } = pkg;
 export function printCreditStatus(remaining, limit = null) {
     if (remaining === null)
@@ -198,16 +198,60 @@ export async function askLoop(engine, options, selectedProvider, selectedModel,
                 if (cmd === "voice") {
                     if (arg === "on") {
                         setVoiceEnabled(true);
-                        console.log(chalk.green("\n  🔊 Voice enabled. Agent responses will be spoken aloud.\n"));
+                        console.log(chalk.green(`\n  🔊 Voice enabled (${getCurrentVoice()} · ${getCurrentTtsModel()}).\n`));
                     }
                     else if (arg === "off") {
                         setVoiceEnabled(false);
                         stopPlayback();
                         console.log(chalk.yellow("\n  🔇 Voice disabled.\n"));
                     }
+                    else if (arg === "list-voices" || arg === "voices") {
+                        console.log(chalk.cyan("\n  Available voices:"));
+                        for (const v of AVAILABLE_VOICES) {
+                            const active = v === getCurrentVoice() ? chalk.green(" ← active") : "";
+                            console.log(chalk.dim(`    ${v}${active}`));
+                        }
+                        console.log(chalk.dim("\n  Usage: /voice set-voice Puck\n"));
+                    }
+                    else if (arg.startsWith("set-voice ")) {
+                        const name = arg.slice("set-voice ".length).trim();
+                        const match = AVAILABLE_VOICES.find(v => v.toLowerCase() === name.toLowerCase());
+                        if (!match) {
+                            console.log(chalk.red(`\n  Unknown voice: "${name}". Run /voice list-voices to see options.\n`));
+                        }
+                        else {
+                            setCurrentVoice(match);
+                            console.log(chalk.green(`\n  🎵 Voice set to ${chalk.bold(match)}.\n`));
+                        }
+                    }
+                    else if (arg === "list-models" || arg === "models") {
+                        console.log(chalk.cyan("\n  Available TTS models:"));
+                        for (const m of AVAILABLE_TTS_MODELS) {
+                            const active = m === getCurrentTtsModel() ? chalk.green(" ← active") : "";
+                            console.log(chalk.dim(`    ${m}${active}`));
+                        }
+                        console.log(chalk.dim("\n  Usage: /voice set-model gemini-2.5-pro-preview-tts\n"));
+                    }
+                    else if (arg.startsWith("set-model ")) {
+                        const name = arg.slice("set-model ".length).trim();
+                        const match = AVAILABLE_TTS_MODELS.find(m => m === name);
+                        if (!match) {
+                            console.log(chalk.red(`\n  Unknown model: "${name}". Run /voice list-models to see options.\n`));
+                        }
+                        else {
+                            setCurrentTtsModel(match);
+                            console.log(chalk.green(`\n  ⚙️  TTS model set to ${chalk.bold(match)}.\n`));
+                        }
+                    }
                     else {
                         const status = isVoiceEnabled() ? chalk.green("on") : chalk.yellow("off");
-                        console.log(chalk.dim(`\n  Voice is currently ${status}. Usage: /voice on  or  /voice off\n`));
+                        console.log(chalk.dim(`\n  Voice is ${status}  ·  Voice: ${chalk.white(getCurrentVoice())}  ·  Model: ${chalk.white(getCurrentTtsModel())}`));
+                        console.log(chalk.dim("\n  Commands:"));
+                        console.log(chalk.dim("    /voice on | off"));
+                        console.log(chalk.dim("    /voice set-voice <name>   e.g. /voice set-voice Puck"));
+                        console.log(chalk.dim("    /voice list-voices"));
+                        console.log(chalk.dim("    /voice set-model <name>   e.g. /voice set-model gemini-2.5-pro-preview-tts"));
+                        console.log(chalk.dim("    /voice list-models\n"));
                     }
                     return ask();
                 }
@@ -217,8 +261,7 @@ export async function askLoop(engine, options, selectedProvider, selectedModel,
                         console.log(chalk.dim("\n  Nothing to speak yet. Ask the agent something first.\n"));
                     }
                     else {
-                        const geminiKey = getGeminiKey() || process.env.GEMINI_API_KEY;
-                        speakText(last, geminiKey ?? undefined).catch(() => { });
+                        speakText(last).catch(() => { });
                         console.log(chalk.dim("\n  🔊 Speaking last response...\n"));
                     }
                     return ask();
@@ -525,8 +568,7 @@ export async function askLoop(engine, options, selectedProvider, selectedModel,
                 if (responseAccumulator) {
                     setLastResponse(responseAccumulator);
                     if (isVoiceEnabled()) {
-                        const geminiKey = getGeminiKey() || process.env.GEMINI_API_KEY;
-                        speakText(responseAccumulator, geminiKey ?? undefined).catch(() => { });
+                        speakText(responseAccumulator).catch(() => { });
                     }
                 }
                 // ── Clean turn separator after every AI reply ─────────────────────────────

package/dist/lib/tts.d.ts

@@ -1,26 +1,27 @@
 /**
  * tts.ts — Text-to-Speech engine for the CareerVivid REPL
  *
- * Uses the Gemini API (TTS model) to synthesize speech from text,
- * writes the result to a temp WAV file, and plays it via the OS
- * audio player (afplay on macOS, aplay on Linux) without blocking
- * the main thread.
+ * Authenticates using the user's CareerVivid API key (cv_live_...) to fetch
+ * a short-lived Gemini key from the backend — exactly like `cv interview`.
+ * No separate GEMINI_API_KEY required.
  *
  * Toggle:  /voice on  |  /voice off
- * Speak:   /speak      (reads the last agent response)
+ * Replay:  /speak
  */
+export declare const AVAILABLE_VOICES: readonly ["Zephyr", "Puck", "Charon", "Kore", "Fenrir", "Aoede", "Orbit", "Stellar", "Leda", "Orus"];
+export declare const AVAILABLE_TTS_MODELS: readonly ["gemini-2.5-flash-preview-tts", "gemini-2.5-pro-preview-tts"];
 export declare function isVoiceEnabled(): boolean;
 export declare function setVoiceEnabled(on: boolean): void;
 export declare function setLastResponse(text: string): void;
 export declare function getLastResponse(): string;
-/**
- * Stops any currently playing audio immediately.
- */
+export declare function getCurrentVoice(): string;
+export declare function setCurrentVoice(v: string): void;
+export declare function getCurrentTtsModel(): string;
+export declare function setCurrentTtsModel(m: string): void;
 export declare function stopPlayback(): void;
 /**
- * Synthesizes `text` via Gemini TTS and plays it asynchronously.
- * Does nothing if voice is disabled. Silently ignores errors so the
- * REPL is never disrupted by TTS failures.
+ * Synthesizes `text` via Gemini TTS using the CareerVivid API key for auth.
+ * Non-blocking — errors are silently swallowed so the REPL is never disrupted.
  */
-export declare function speakText(text: string, apiKey?: string): Promise<void>;
+export declare function speakText(text: string, _unusedKey?: string): Promise<void>;
 //# sourceMappingURL=tts.d.ts.map

package/dist/lib/tts.d.ts.map

@@ -1 +1 @@
-{"version":3,"file":"tts.d.ts","sourceRoot":"","sources":["../../src/lib/tts.ts"],"names":[],"mappings":"AAAA;;;;;;;;;;GAUG;AAaH,wBAAgB,cAAc,YAA2B;AACzD,wBAAgB,eAAe,CAAC,EAAE,EAAE,OAAO,QAAwB;AACnE,wBAAgB,eAAe,CAAC,IAAI,EAAE,MAAM,QAA0B;AACtE,wBAAgB,eAAe,WAA2B;AAI1D;;GAEG;AACH,wBAAgB,YAAY,SAK3B;AAoED;;;;GAIG;AACH,wBAAsB,SAAS,CAAC,IAAI,EAAE,MAAM,EAAE,MAAM,CAAC,EAAE,MAAM,GAAG,OAAO,CAAC,IAAI,CAAC,CAuD5E"}
+{"version":3,"file":"tts.d.ts","sourceRoot":"","sources":["../../src/lib/tts.ts"],"names":[],"mappings":"AAAA;;;;;;;;;GASG;AAgBH,eAAO,MAAM,gBAAgB,sGAWnB,CAAC;AAEX,eAAO,MAAM,oBAAoB,yEAGvB,CAAC;AAYX,wBAAgB,cAAc,YAA2B;AACzD,wBAAgB,eAAe,CAAC,EAAE,EAAE,OAAO,QAAwB;AACnE,wBAAgB,eAAe,CAAC,IAAI,EAAE,MAAM,QAA0B;AACtE,wBAAgB,eAAe,WAA2B;AAC1D,wBAAgB,eAAe,WAA2B;AAC1D,wBAAgB,eAAe,CAAC,CAAC,EAAE,MAAM,QAAuB;AAChE,wBAAgB,kBAAkB,WAA8B;AAChE,wBAAgB,kBAAkB,CAAC,CAAC,EAAE,MAAM,QAA0B;AA8BtE,wBAAgB,YAAY,SAK3B;AA2DD;;;GAGG;AACH,wBAAsB,SAAS,CAAC,IAAI,EAAE,MAAM,EAAE,UAAU,CAAC,EAAE,MAAM,GAAG,OAAO,CAAC,IAAI,CAAC,CAuDhF"}

package/dist/lib/tts.js

@@ -1,41 +1,89 @@
 /**
  * tts.ts — Text-to-Speech engine for the CareerVivid REPL
  *
- * Uses the Gemini API (TTS model) to synthesize speech from text,
- * writes the result to a temp WAV file, and plays it via the OS
- * audio player (afplay on macOS, aplay on Linux) without blocking
- * the main thread.
+ * Authenticates using the user's CareerVivid API key (cv_live_...) to fetch
+ * a short-lived Gemini key from the backend — exactly like `cv interview`.
+ * No separate GEMINI_API_KEY required.
  *
  * Toggle:  /voice on  |  /voice off
- * Speak:   /speak      (reads the last agent response)
+ * Replay:  /speak
  */
 import { writeFileSync, unlinkSync } from "fs";
 import { spawn } from "child_process";
 import { tmpdir } from "os";
 import { join } from "path";
 import { GoogleGenAI, Modality } from "@google/genai";
+import { getApiKey } from "../config.js";
+// ── Backend endpoint (same as interview token vend) ───────────────────────────
+const TTS_TOKEN_URL = process.env.CV_FUNCTIONS_URL
+    ? `${process.env.CV_FUNCTIONS_URL}/cliGetInterviewToken`
+    : "https://us-west1-jastalk-firebase.cloudfunctions.net/cliGetInterviewToken";
+// ── Available options ────────────────────────────────────────────────────────
+export const AVAILABLE_VOICES = [
+    "Zephyr", // Bright, energetic
+    "Puck", // Upbeat, playful
+    "Charon", // Informative, measured
+    "Kore", // Firm, confident
+    "Fenrir", // Excitable, dynamic
+    "Aoede", // Breezy, easy-going
+    "Orbit", // Friendly, relaxed
+    "Stellar", // Smooth, polished
+    "Leda", // Warm, natural
+    "Orus", // Confident, authoritative
+];
+export const AVAILABLE_TTS_MODELS = [
+    "gemini-2.5-flash-preview-tts", // Fast, efficient (default)
+    "gemini-2.5-pro-preview-tts", // Higher quality, slower
+];
 // ── State ────────────────────────────────────────────────────────────────────
 let voiceEnabled = false;
 let lastResponse = "";
 let playbackProcess = null;
+let currentVoice = "Zephyr";
+let currentTtsModel = "gemini-2.5-flash-preview-tts";
+// Cache the Gemini key for the session so we don't hit the endpoint every turn
+let cachedGeminiKey = null;
 export function isVoiceEnabled() { return voiceEnabled; }
 export function setVoiceEnabled(on) { voiceEnabled = on; }
 export function setLastResponse(text) { lastResponse = text; }
 export function getLastResponse() { return lastResponse; }
+export function getCurrentVoice() { return currentVoice; }
+export function setCurrentVoice(v) { currentVoice = v; }
+export function getCurrentTtsModel() { return currentTtsModel; }
+export function setCurrentTtsModel(m) { currentTtsModel = m; }
+// ── Gemini key via CV API key ─────────────────────────────────────────────────
+async function fetchGeminiKey() {
+    if (cachedGeminiKey)
+        return cachedGeminiKey;
+    const apiKey = getApiKey();
+    if (!apiKey)
+        return null;
+    try {
+        const res = await fetch(TTS_TOKEN_URL, {
+            method: "POST",
+            headers: { "Content-Type": "application/json" },
+            body: JSON.stringify({ apiKey, role: "tts" }),
+        });
+        if (!res.ok)
+            return null;
+        const data = await res.json();
+        if (data?.geminiKey) {
+            cachedGeminiKey = data.geminiKey;
+            return cachedGeminiKey;
+        }
+    }
+    catch {
+        // Network error — fall through
+    }
+    return null;
+}
 // ── Audio Playback ────────────────────────────────────────────────────────────
-/**
- * Stops any currently playing audio immediately.
- */
 export function stopPlayback() {
     if (playbackProcess && !playbackProcess.killed) {
         playbackProcess.kill("SIGKILL");
         playbackProcess = null;
     }
 }
-/**
- * Plays a WAV buffer using the OS audio player (non-blocking).
- * afplay on macOS, aplay on Linux, powershell on Windows.
- */
 function playWav(wavBuffer) {
     const tmpFile = join(tmpdir(), `cv-tts-${Date.now()}.wav`);
     writeFileSync(tmpFile, wavBuffer);
@@ -51,30 +99,28 @@ function playWav(wavBuffer) {
         playerArgs = ["-q", tmpFile];
     }
     else {
-        // Windows fallback via PowerShell
         playerCmd = "powershell";
         playerArgs = ["-c", `(New-Object System.Media.SoundPlayer '${tmpFile}').PlaySync()`];
     }
-    stopPlayback(); // Stop any previous playback
+    stopPlayback();
     const child = spawn(playerCmd, playerArgs, { stdio: "ignore", detached: false });
     playbackProcess = child;
     child.on("close", () => {
         try {
             unlinkSync(tmpFile);
         }
-        catch { /* ignore cleanup errors */ }
+        catch { /* ignore */ }
         if (playbackProcess === child)
             playbackProcess = null;
     });
     child.on("error", () => {
-        // Silently ignore — player may not be installed
         try {
             unlinkSync(tmpFile);
         }
         catch { /* ignore */ }
     });
 }
-// ── WAV Header Builder ────────────────────────────────────────────────────────
+// ── WAV Builder ───────────────────────────────────────────────────────────────
 function buildWavHeader(dataLength, sampleRate = 24000, channels = 1, bitsPerSample = 16) {
     const byteRate = sampleRate * channels * bitsPerSample / 8;
     const blockAlign = channels * bitsPerSample / 8;
@@ -94,40 +140,39 @@ function buildWavHeader(dataLength, sampleRate = 24000, channels = 1, bitsPerSam
     header.writeUInt32LE(dataLength, 40);
     return header;
 }
-// ── TTS Synthesis ────────────────────────────────────────────────────────────
+// ── TTS Synthesis ─────────────────────────────────────────────────────────────
 /**
- * Synthesizes `text` via Gemini TTS and plays it asynchronously.
- * Does nothing if voice is disabled. Silently ignores errors so the
- * REPL is never disrupted by TTS failures.
+ * Synthesizes `text` via Gemini TTS using the CareerVivid API key for auth.
+ * Non-blocking — errors are silently swallowed so the REPL is never disrupted.
  */
-export async function speakText(text, apiKey) {
+export async function speakText(text, _unusedKey) {
     if (!text.trim())
         return;
-    const key = apiKey || process.env.GEMINI_API_KEY;
-    if (!key)
-        return; // No key — silently skip
-    // Strip markdown so the audio sounds natural
+    const geminiKey = await fetchGeminiKey();
+    if (!geminiKey)
+        return; // No key available — silently skip
+    // Strip markdown for natural-sounding speech
     const cleaned = text
-        .replace(/```[\s\S]*?```/g, "") // code blocks
-        .replace(/`[^`]+`/g, "") // inline code
-        .replace(/\*\*(.*?)\*\*/g, "$1") // bold
-        .replace(/\*(.*?)\*/g, "$1") // italic
-        .replace(/^[#>•\-*]\s*/gm, "") // headings, quotes, bullets
+        .replace(/```[\s\S]*?```/g, "")
+        .replace(/`[^`]+`/g, "")
+        .replace(/\*\*(.*?)\*\*/g, "$1")
+        .replace(/\*(.*?)\*/g, "$1")
+        .replace(/^[#>•\-*]\s*/gm, "")
         .replace(/\s+/g, " ")
         .trim()
-        .slice(0, 1000); // Cap to ~1000 chars to keep TTS fast
+        .slice(0, 1000);
     if (!cleaned)
         return;
     try {
-        const ai = new GoogleGenAI({ apiKey: key });
+        const ai = new GoogleGenAI({ apiKey: geminiKey });
         const response = await ai.models.generateContent({
-            model: "gemini-2.5-flash-preview-tts",
+            model: currentTtsModel,
             contents: [{ parts: [{ text: cleaned }] }],
             config: {
                 responseModalities: [Modality.AUDIO],
                 speechConfig: {
                     voiceConfig: {
-                        prebuiltVoiceConfig: { voiceName: "Zephyr" },
+                        prebuiltVoiceConfig: { voiceName: currentVoice },
                     },
                 },
             },
@@ -142,12 +187,12 @@ export async function speakText(text, apiKey) {
         if (audioParts.length === 0)
             return;
         const pcmData = Buffer.concat(audioParts);
-        const wavHeader = buildWavHeader(pcmData.length);
-        const wavBuffer = Buffer.concat([wavHeader, pcmData]);
-        // Fire and forget — does not block the REPL
+        const wavBuffer = Buffer.concat([buildWavHeader(pcmData.length), pcmData]);
         playWav(wavBuffer);
     }
     catch {
-        // Silently swallow TTS errors — never disrupt the agent session
+        // Silently ignore — TTS errors must never crash the agent REPL
+        // Invalidate cached key so we retry fetching on the next call
+        cachedGeminiKey = null;
     }
 }

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "careervivid",
-  "version": "2.1.13",
+  "version": "2.1.16",
   "description": "Official CLI for CareerVivid — AI voice interviews, autonomous job applications, resume editing, and portfolio publishing from your terminal",
   "type": "module",
   "bin": {