agentgui 1.0.143 → 1.0.145

package/database.js

@@ -510,6 +510,13 @@ export const queries = {
     return stmt.all(conversationId, status);
   },
 
+  getActiveSessions() {
+    const stmt = db.prepare(
+      "SELECT * FROM sessions WHERE status IN ('active', 'pending') ORDER BY started_at DESC"
+    );
+    return stmt.all();
+  },
+
   createEvent(type, data, conversationId = null, sessionId = null) {
     const id = generateId('evt');
     const now = Date.now();

package/lib/claude-runner.js

@@ -52,6 +52,10 @@ class AgentRunner {
       const args = this.buildArgs(prompt, config);
       const proc = spawn(this.command, args, { cwd });
 
+      if (config.onPid) {
+        try { config.onPid(proc.pid); } catch (e) {}
+      }
+
       let jsonBuffer = '';
       const outputs = [];
       let timedOut = false;
@@ -150,6 +154,10 @@ class AgentRunner {
       
       const proc = spawn(cmd, args, { cwd });
 
+      if (config.onPid) {
+        try { config.onPid(proc.pid); } catch (e) {}
+      }
+
       const outputs = [];
       let timedOut = false;
       let sessionId = null;

package/lib/speech.js

@@ -0,0 +1,192 @@
+import { pipeline, env } from '@huggingface/transformers';
+import { createRequire } from 'module';
+import fs from 'fs';
+import path from 'path';
+import { fileURLToPath } from 'url';
+
+const require = createRequire(import.meta.url);
+const __dirname = path.dirname(fileURLToPath(import.meta.url));
+const ROOT = path.dirname(__dirname);
+const DATA_DIR = path.join(ROOT, 'data');
+
+const SPEAKER_EMBEDDINGS_URL = 'https://huggingface.co/datasets/Xenova/transformers.js-docs/resolve/main/speaker_embeddings.bin';
+const SPEAKER_EMBEDDINGS_PATH = path.join(DATA_DIR, 'speaker_embeddings.bin');
+const SAMPLE_RATE_TTS = 16000;
+const SAMPLE_RATE_STT = 16000;
+
+let sttPipeline = null;
+let ttsPipeline = null;
+let speakerEmbeddings = null;
+let sttLoading = false;
+let ttsLoading = false;
+
+function whisperModelPath() {
+  try {
+    const webtalkDir = path.dirname(require.resolve('webtalk'));
+    const p = path.join(webtalkDir, 'models', 'onnx-community', 'whisper-base');
+    if (fs.existsSync(p)) return p;
+  } catch (_) {}
+  return 'onnx-community/whisper-base';
+}
+
+async function ensureSpeakerEmbeddings() {
+  if (speakerEmbeddings) return speakerEmbeddings;
+  if (!fs.existsSync(DATA_DIR)) fs.mkdirSync(DATA_DIR, { recursive: true });
+  if (!fs.existsSync(SPEAKER_EMBEDDINGS_PATH)) {
+    const resp = await fetch(SPEAKER_EMBEDDINGS_URL);
+    if (!resp.ok) throw new Error('Failed to download speaker embeddings');
+    fs.writeFileSync(SPEAKER_EMBEDDINGS_PATH, Buffer.from(await resp.arrayBuffer()));
+  }
+  const buf = fs.readFileSync(SPEAKER_EMBEDDINGS_PATH);
+  speakerEmbeddings = new Float32Array(new Uint8Array(buf).buffer);
+  return speakerEmbeddings;
+}
+
+async function getSTT() {
+  if (sttPipeline) return sttPipeline;
+  if (sttLoading) {
+    while (sttLoading) await new Promise(r => setTimeout(r, 100));
+    return sttPipeline;
+  }
+  sttLoading = true;
+  try {
+    const modelPath = whisperModelPath();
+    const isLocal = !modelPath.includes('/') || fs.existsSync(modelPath);
+    env.allowLocalModels = true;
+    env.allowRemoteModels = !isLocal;
+    if (isLocal) env.localModelPath = '';
+    sttPipeline = await pipeline('automatic-speech-recognition', modelPath, {
+      device: 'cpu',
+      local_files_only: isLocal,
+    });
+    return sttPipeline;
+  } finally {
+    sttLoading = false;
+  }
+}
+
+async function getTTS() {
+  if (ttsPipeline) return ttsPipeline;
+  if (ttsLoading) {
+    while (ttsLoading) await new Promise(r => setTimeout(r, 100));
+    return ttsPipeline;
+  }
+  ttsLoading = true;
+  try {
+    env.allowRemoteModels = true;
+    ttsPipeline = await pipeline('text-to-speech', 'Xenova/speecht5_tts', {
+      device: 'cpu',
+      dtype: 'fp32',
+    });
+    await ensureSpeakerEmbeddings();
+    return ttsPipeline;
+  } finally {
+    ttsLoading = false;
+  }
+}
+
+function decodeWavToFloat32(buffer) {
+  const view = new DataView(buffer.buffer || buffer);
+  const riff = String.fromCharCode(view.getUint8(0), view.getUint8(1), view.getUint8(2), view.getUint8(3));
+  if (riff !== 'RIFF') throw new Error('Not a WAV file');
+  const numChannels = view.getUint16(22, true);
+  const sampleRate = view.getUint32(24, true);
+  const bitsPerSample = view.getUint16(34, true);
+  let dataOffset = 44;
+  for (let i = 36; i < view.byteLength - 8; i++) {
+    if (view.getUint8(i) === 0x64 && view.getUint8(i+1) === 0x61 &&
+        view.getUint8(i+2) === 0x74 && view.getUint8(i+3) === 0x61) {
+      dataOffset = i + 8;
+      break;
+    }
+  }
+  const bytesPerSample = bitsPerSample / 8;
+  const numSamples = Math.floor((view.byteLength - dataOffset) / (bytesPerSample * numChannels));
+  const audio = new Float32Array(numSamples);
+  for (let i = 0; i < numSamples; i++) {
+    const offset = dataOffset + i * bytesPerSample * numChannels;
+    if (bitsPerSample === 16) {
+      audio[i] = view.getInt16(offset, true) / 32768;
+    } else if (bitsPerSample === 32) {
+      audio[i] = view.getFloat32(offset, true);
+    } else {
+      audio[i] = (view.getUint8(offset) - 128) / 128;
+    }
+  }
+  return { audio, sampleRate };
+}
+
+function resampleTo16k(audio, fromRate) {
+  if (fromRate === SAMPLE_RATE_STT) return audio;
+  const ratio = fromRate / SAMPLE_RATE_STT;
+  const newLen = Math.round(audio.length / ratio);
+  const result = new Float32Array(newLen);
+  for (let i = 0; i < newLen; i++) {
+    const srcIdx = i * ratio;
+    const lo = Math.floor(srcIdx);
+    const hi = Math.min(lo + 1, audio.length - 1);
+    const frac = srcIdx - lo;
+    result[i] = audio[lo] * (1 - frac) + audio[hi] * frac;
+  }
+  return result;
+}
+
+function encodeWav(float32Audio, sampleRate) {
+  const numSamples = float32Audio.length;
+  const bytesPerSample = 2;
+  const dataSize = numSamples * bytesPerSample;
+  const buffer = new ArrayBuffer(44 + dataSize);
+  const view = new DataView(buffer);
+  const writeStr = (off, str) => { for (let i = 0; i < str.length; i++) view.setUint8(off + i, str.charCodeAt(i)); };
+  writeStr(0, 'RIFF');
+  view.setUint32(4, 36 + dataSize, true);
+  writeStr(8, 'WAVE');
+  writeStr(12, 'fmt ');
+  view.setUint32(16, 16, true);
+  view.setUint16(20, 1, true);
+  view.setUint16(22, 1, true);
+  view.setUint32(24, sampleRate, true);
+  view.setUint32(28, sampleRate * bytesPerSample, true);
+  view.setUint16(32, bytesPerSample, true);
+  view.setUint16(34, 16, true);
+  writeStr(36, 'data');
+  view.setUint32(40, dataSize, true);
+  for (let i = 0; i < numSamples; i++) {
+    const s = Math.max(-1, Math.min(1, float32Audio[i]));
+    view.setInt16(44 + i * 2, s < 0 ? s * 32768 : s * 32767, true);
+  }
+  return Buffer.from(buffer);
+}
+
+async function transcribe(audioBuffer) {
+  const stt = await getSTT();
+  let audio;
+  const buf = Buffer.isBuffer(audioBuffer) ? audioBuffer : Buffer.from(audioBuffer);
+  const isWav = buf.length > 4 && buf.toString('ascii', 0, 4) === 'RIFF';
+  if (isWav) {
+    const decoded = decodeWavToFloat32(buf);
+    audio = resampleTo16k(decoded.audio, decoded.sampleRate);
+  } else {
+    audio = new Float32Array(buf.buffer, buf.byteOffset, buf.byteLength / 4);
+  }
+  const result = await stt(audio);
+  return result.text || '';
+}
+
+async function synthesize(text) {
+  const tts = await getTTS();
+  const embeddings = await ensureSpeakerEmbeddings();
+  const result = await tts(text, { speaker_embeddings: embeddings });
+  return encodeWav(result.audio, result.sampling_rate || SAMPLE_RATE_TTS);
+}
+
+function getStatus() {
+  return {
+    sttReady: !!sttPipeline,
+    ttsReady: !!ttsPipeline,
+    sttLoading,
+    ttsLoading,
+  };
+}
+
+export { transcribe, synthesize, getSTT, getTTS, getStatus };

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "agentgui",
-  "version": "1.0.143",
+  "version": "1.0.145",
   "description": "Multi-agent ACP client with real-time communication",
   "type": "module",
   "main": "server.js",
@@ -22,10 +22,12 @@
   },
   "dependencies": {
     "@anthropic-ai/claude-code": "^2.1.37",
+    "@huggingface/transformers": "^3.8.1",
     "better-sqlite3": "^12.6.2",
     "busboy": "^1.6.0",
     "express": "^5.2.1",
     "fsbrowse": "^0.2.13",
+    "onnxruntime-node": "^1.24.1",
     "webtalk": "github:anEntrypoint/realtime-whisper-webgpu",
     "ws": "^8.14.2"
   }

package/server.js

@@ -8,12 +8,12 @@ import { execSync } from 'child_process';
 import { createRequire } from 'module';
 import { queries } from './database.js';
 import { runClaudeWithStreaming } from './lib/claude-runner.js';
+import { transcribe, synthesize, getStatus as getSpeechStatus } from './lib/speech.js';
 
 const require = createRequire(import.meta.url);
 const express = require('express');
 const Busboy = require('busboy');
 const fsbrowse = require('fsbrowse');
-const { webtalk } = require('webtalk');
 
 const SYSTEM_PROMPT = `Always write your responses in ripple-ui enhanced HTML. Avoid overriding light/dark mode CSS variables. Use all the benefits of HTML to express technical details with proper semantic markup, tables, code blocks, headings, and lists. Write clean, well-structured HTML that respects the existing design system.`;
 
@@ -37,28 +37,6 @@ if (!fs.existsSync(staticDir)) fs.mkdirSync(staticDir, { recursive: true });
 // Express sub-app for fsbrowse file browser and file upload
 const expressApp = express();
 
-// Separate Express app for webtalk (STT/TTS) - isolated to contain COEP/COOP headers
-const webtalkApp = express();
-const webtalkInstance = webtalk(webtalkApp, { path: '/webtalk' });
-
-const webtalkSdkDir = path.dirname(require.resolve('webtalk'));
-const WASM_MIN_BYTES = 1000000;
-const webtalkCriticalFiles = [
-  { path: path.join(webtalkSdkDir, 'assets', 'ort-wasm-simd-threaded.jsep.wasm'), minBytes: WASM_MIN_BYTES }
-];
-for (const file of webtalkCriticalFiles) {
-  try {
-    if (fs.existsSync(file.path)) {
-      const stat = fs.statSync(file.path);
-      if (stat.size < file.minBytes) {
-        debugLog(`Removing corrupt file ${path.basename(file.path)} (${stat.size} bytes, need ${file.minBytes}+)`);
-        fs.unlinkSync(file.path);
-      }
-    }
-  } catch (e) { debugLog(`File check error: ${e.message}`); }
-}
-
-webtalkInstance.init().catch(err => debugLog('Webtalk init: ' + err.message));
 
 // File upload endpoint - copies dropped files to conversation workingDirectory
 expressApp.post(BASE_URL + '/api/upload/:conversationId', (req, res) => {
@@ -165,63 +143,9 @@ const server = http.createServer(async (req, res) => {
   res.setHeader('Access-Control-Allow-Origin', '*');
   res.setHeader('Access-Control-Allow-Methods', 'GET, POST, PUT, DELETE, OPTIONS');
   res.setHeader('Access-Control-Allow-Headers', 'Content-Type, Authorization');
-  res.setHeader('Cross-Origin-Embedder-Policy', 'credentialless');
-  res.setHeader('Cross-Origin-Opener-Policy', 'same-origin');
-  res.setHeader('Cross-Origin-Resource-Policy', 'cross-origin');
   if (req.method === 'OPTIONS') { res.writeHead(200); res.end(); return; }
 
   const pathOnly = req.url.split('?')[0];
-  const webtalkPrefix = BASE_URL + '/webtalk';
-  const isWebtalkRoute = pathOnly.startsWith(webtalkPrefix) ||
-    pathOnly.startsWith(BASE_URL + '/api/tts-status') ||
-    pathOnly.startsWith(BASE_URL + '/assets/') ||
-    pathOnly.startsWith(BASE_URL + '/tts/') ||
-    pathOnly.startsWith(BASE_URL + '/models/') ||
-    pathOnly.startsWith('/webtalk') ||
-    pathOnly.startsWith('/assets/') ||
-    pathOnly.startsWith('/tts/') ||
-    pathOnly.startsWith('/models/');
-  if (isWebtalkRoute) {
-    const webtalkSdkDir = path.dirname(require.resolve('webtalk'));
-    const sdkFiles = { '/demo': 'app.html', '/sdk.js': 'sdk.js', '/stt.js': 'stt.js', '/tts.js': 'tts.js', '/tts-utils.js': 'tts-utils.js' };
-    let stripped = pathOnly.startsWith(webtalkPrefix) ? pathOnly.slice(webtalkPrefix.length) : (pathOnly.startsWith('/webtalk') ? pathOnly.slice('/webtalk'.length) : null);
-    if (stripped !== null && !sdkFiles[stripped] && !stripped.endsWith('.js') && sdkFiles[stripped + '.js']) stripped += '.js';
-    if (stripped !== null && sdkFiles[stripped]) {
-      const filePath = path.join(webtalkSdkDir, sdkFiles[stripped]);
-      return fs.readFile(filePath, 'utf-8', (err, content) => {
-        if (err) { res.writeHead(404); res.end('Not found'); return; }
-        if (stripped === '/demo') {
-          let patched = content
-            .replace(/from\s+['"](\/webtalk\/[^'"]+)['"]/g, (_, p) => `from '${BASE_URL}${p}'`)
-            .replace(/from\s+['"]\.\/([^'"]+)['"]/g, (_, p) => `from '${BASE_URL}/webtalk/${p}'`)
-            .replace('<head>', `<head>\n    <script>window.__WEBTALK_BASE='${BASE_URL}';</script>`);
-          res.writeHead(200, { 'Content-Type': 'text/html; charset=utf-8', 'Cross-Origin-Embedder-Policy': 'credentialless', 'Cross-Origin-Opener-Policy': 'same-origin', 'Cross-Origin-Resource-Policy': 'cross-origin' });
-          return res.end(patched);
-        }
-        let js = content;
-        const ensureExt = (mod) => mod.endsWith('.js') ? mod : mod + '.js';
-        if (js.includes('require(') || js.includes('module.exports')) {
-          js = js.replace(/const\s*\{([^}]+)\}\s*=\s*require\(['"]\.\/([^'"]+)['"]\);?/g, (_, names, mod) => `import {${names}} from '${BASE_URL}/webtalk/${ensureExt(mod)}';`);
-          js = js.replace(/const\s+(\w+)\s*=\s*require\(['"]\.\/([^'"]+)['"]\);?/g, (_, name, mod) => `import ${name} from '${BASE_URL}/webtalk/${ensureExt(mod)}';`);
-          js = js.replace(/module\.exports\s*=\s*\{([^}]+)\};?/, (_, names) => `export {${names.trim().replace(/\s+/g, ' ')} };`);
-        }
-        js = js.replace(/from\s+['"]\.\/([^'"]+)['"]/g, (_, p) => `from '${BASE_URL}/webtalk/${ensureExt(p)}'`);
-        res.writeHead(200, { 'Content-Type': 'application/javascript; charset=utf-8', 'Cross-Origin-Resource-Policy': 'cross-origin' });
-        res.end(js);
-      });
-    }
-    if (req.url.startsWith(BASE_URL)) req.url = req.url.slice(BASE_URL.length) || '/';
-    const isModelOrAsset = pathOnly.includes('/models/') || pathOnly.includes('/assets/') || pathOnly.endsWith('.wasm') || pathOnly.endsWith('.onnx');
-    if (isModelOrAsset) {
-      res.setHeader('Cache-Control', 'public, max-age=604800, immutable');
-    }
-    const origSetHeader = res.setHeader.bind(res);
-    res.setHeader = (name, value) => {
-      if (name.toLowerCase() === 'cross-origin-embedder-policy') return;
-      origSetHeader(name, value);
-    };
-    return webtalkApp(req, res);
-  }
 
   // Route file upload and fsbrowse requests through Express sub-app
   if (pathOnly.startsWith(BASE_URL + '/api/upload/') || pathOnly.startsWith(BASE_URL + '/files/')) {
@@ -516,6 +440,53 @@ const server = http.createServer(async (req, res) => {
       return;
     }
 
+    if (routePath === '/api/stt' && req.method === 'POST') {
+      try {
+        const chunks = [];
+        for await (const chunk of req) chunks.push(chunk);
+        const audioBuffer = Buffer.concat(chunks);
+        if (audioBuffer.length === 0) {
+          res.writeHead(400, { 'Content-Type': 'application/json' });
+          res.end(JSON.stringify({ error: 'No audio data' }));
+          return;
+        }
+        const text = await transcribe(audioBuffer);
+        res.writeHead(200, { 'Content-Type': 'application/json' });
+        res.end(JSON.stringify({ text: text.trim() }));
+      } catch (err) {
+        debugLog('[STT] Error: ' + err.message);
+        res.writeHead(500, { 'Content-Type': 'application/json' });
+        res.end(JSON.stringify({ error: err.message }));
+      }
+      return;
+    }
+
+    if (routePath === '/api/tts' && req.method === 'POST') {
+      try {
+        const body = await parseBody(req);
+        const text = body.text || '';
+        if (!text) {
+          res.writeHead(400, { 'Content-Type': 'application/json' });
+          res.end(JSON.stringify({ error: 'No text provided' }));
+          return;
+        }
+        const wavBuffer = await synthesize(text);
+        res.writeHead(200, { 'Content-Type': 'audio/wav', 'Content-Length': wavBuffer.length });
+        res.end(wavBuffer);
+      } catch (err) {
+        debugLog('[TTS] Error: ' + err.message);
+        res.writeHead(500, { 'Content-Type': 'application/json' });
+        res.end(JSON.stringify({ error: err.message }));
+      }
+      return;
+    }
+
+    if (routePath === '/api/speech-status' && req.method === 'GET') {
+      res.writeHead(200, { 'Content-Type': 'application/json' });
+      res.end(JSON.stringify(getSpeechStatus()));
+      return;
+    }
+
     if (routePath === '/api/folders' && req.method === 'POST') {
       const body = await parseBody(req);
       const folderPath = body.path || STARTUP_CWD;
@@ -591,7 +562,7 @@ const server = http.createServer(async (req, res) => {
   }
 });
 
-const MIME_TYPES = { '.html': 'text/html; charset=utf-8', '.js': 'application/javascript; charset=utf-8', '.css': 'text/css; charset=utf-8', '.json': 'application/json', '.png': 'image/png', '.jpg': 'image/jpeg', '.svg': 'image/svg+xml', '.wasm': 'application/wasm', '.onnx': 'application/octet-stream' };
+const MIME_TYPES = { '.html': 'text/html; charset=utf-8', '.js': 'application/javascript; charset=utf-8', '.css': 'text/css; charset=utf-8', '.json': 'application/json', '.png': 'image/png', '.jpg': 'image/jpeg', '.svg': 'image/svg+xml' };
 
 function serveFile(filePath, res) {
   const ext = path.extname(filePath).toLowerCase();
@@ -613,7 +584,7 @@ function serveFile(filePath, res) {
   fs.readFile(filePath, (err, data) => {
     if (err) { res.writeHead(500); res.end('Server error'); return; }
     let content = data.toString();
-    const baseTag = `<script>window.__BASE_URL='${BASE_URL}';</script>\n  <script type="importmap">{"imports":{"webtalk-sdk":"${BASE_URL}/webtalk/sdk.js"}}</script>`;
+    const baseTag = `<script>window.__BASE_URL='${BASE_URL}';</script>`;
     content = content.replace('<head>', '<head>\n  ' + baseTag);
     if (watch) {
       content += `\n<script>(function(){const ws=new WebSocket('ws://'+location.host+'${BASE_URL}/hot-reload');ws.onmessage=e=>{if(JSON.parse(e.data).type==='reload')location.reload()};})();</script>`;
@@ -640,7 +611,7 @@ function persistChunkWithRetry(sessionId, conversationId, sequence, blockType, b
 
 async function processMessageWithStreaming(conversationId, messageId, sessionId, content, agentId) {
   const startTime = Date.now();
-  activeExecutions.set(conversationId, true);
+  activeExecutions.set(conversationId, { pid: null, startTime, sessionId });
   queries.setIsStreaming(conversationId, true);
   queries.updateSession(sessionId, { status: 'active' });
 
@@ -756,7 +727,11 @@ async function processMessageWithStreaming(conversationId, messageId, sessionId,
       print: true,
       resumeSessionId,
       systemPrompt: SYSTEM_PROMPT,
-      onEvent
+      onEvent,
+      onPid: (pid) => {
+        const entry = activeExecutions.get(conversationId);
+        if (entry) entry.pid = pid;
+      }
     };
 
     const { outputs, sessionId: claudeSessionId } = await runClaudeWithStreaming(content, cwd, agentId || 'claude-code', config);
@@ -1030,6 +1005,66 @@ server.on('error', (err) => {
   }
 });
 
+function recoverStaleSessions() {
+  try {
+    const staleSessions = queries.getActiveSessions ? queries.getActiveSessions() : [];
+    let recoveredCount = 0;
+    for (const session of staleSessions) {
+      if (!activeExecutions.has(session.conversationId)) {
+        queries.updateSession(session.id, {
+          status: 'error',
+          error: 'Agent died unexpectedly (server restart)',
+          completed_at: Date.now()
+        });
+        queries.setIsStreaming(session.conversationId, false);
+        broadcastSync({
+          type: 'streaming_error',
+          sessionId: session.id,
+          conversationId: session.conversationId,
+          error: 'Agent died unexpectedly (server restart)',
+          recoverable: false,
+          timestamp: Date.now()
+        });
+        recoveredCount++;
+      }
+    }
+    if (recoveredCount > 0) {
+      console.log(`[RECOVERY] Recovered ${recoveredCount} stale active session(s)`);
+    }
+  } catch (err) {
+    console.error('[RECOVERY] Stale session recovery error:', err.message);
+  }
+}
+
+function performAgentHealthCheck() {
+  for (const [conversationId, entry] of activeExecutions) {
+    if (!entry || !entry.pid) continue;
+    try {
+      process.kill(entry.pid, 0);
+    } catch (err) {
+      debugLog(`[HEALTH] Agent PID ${entry.pid} for conv ${conversationId} is dead`);
+      activeExecutions.delete(conversationId);
+      queries.setIsStreaming(conversationId, false);
+      if (entry.sessionId) {
+        queries.updateSession(entry.sessionId, {
+          status: 'error',
+          error: 'Agent process died unexpectedly',
+          completed_at: Date.now()
+        });
+      }
+      broadcastSync({
+        type: 'streaming_error',
+        sessionId: entry.sessionId,
+        conversationId,
+        error: 'Agent process died unexpectedly',
+        recoverable: false,
+        timestamp: Date.now()
+      });
+      drainMessageQueue(conversationId);
+    }
+  }
+}
+
 function onServerReady() {
   console.log(`GMGUI running on http://localhost:${PORT}${BASE_URL}/`);
   console.log(`Agents: ${discoveredAgents.map(a => a.name).join(', ') || 'none'}`);
@@ -1041,12 +1076,18 @@ function onServerReady() {
     console.log(`Cleaned up ${deletedCount} empty conversation(s) on startup`);
   }
 
+  // Recover stale active sessions from previous run
+  recoverStaleSessions();
+
   // Run auto-import immediately
   performAutoImport();
 
   // Then run it every 30 seconds (constant automatic importing)
   setInterval(performAutoImport, 30000);
 
+  // Agent health check every 30 seconds
+  setInterval(performAgentHealthCheck, 30000);
+
 }
 
 function performAutoImport() {

package/static/js/client.js

@@ -792,7 +792,13 @@ class AgentGUIClient {
               const inputStr = JSON.stringify(block.input, null, 2);
               inputHtml = `<details class="tool-input-details"><summary class="tool-input-summary">Input</summary><pre class="tool-input-pre">${this.escapeHtml(inputStr)}</pre></details>`;
             }
-            html += `<div class="streaming-block-tool-use"><div class="tool-use-header"><span class="tool-use-icon">&#9881;</span> <span class="tool-use-name">${this.escapeHtml(block.name || 'unknown')}</span></div>${inputHtml}</div>`;
+            const tn = block.name || 'unknown';
+            const foldable = tn.startsWith('mcp__') || tn === 'Edit';
+            if (foldable) {
+              html += `<details class="streaming-block-tool-use"><summary class="tool-use-header" style="cursor:pointer;user-select:none;list-style:none;"><span class="tool-use-icon">&#9881;</span> <span class="tool-use-name">${this.escapeHtml(tn)}</span></summary>${inputHtml}</details>`;
+            } else {
+              html += `<div class="streaming-block-tool-use"><div class="tool-use-header"><span class="tool-use-icon">&#9881;</span> <span class="tool-use-name">${this.escapeHtml(tn)}</span></div>${inputHtml}</div>`;
+            }
           } else if (block.type === 'tool_result') {
             const content = typeof block.content === 'string' ? block.content : JSON.stringify(block.content);
             const smartHtml = typeof StreamingRenderer !== 'undefined' ? StreamingRenderer.renderSmartContentHTML(content, this.escapeHtml.bind(this)) : `<pre class="tool-result-pre">${this.escapeHtml(content.length > 2000 ? content.substring(0, 2000) + '\n... (truncated)' : content)}</pre>`;
@@ -1433,7 +1439,13 @@ class AgentGUIClient {
                 const inputStr = JSON.stringify(block.input, null, 2);
                 inputHtml = `<details class="tool-input-details"><summary class="tool-input-summary">Input</summary><pre class="tool-input-pre">${this.escapeHtml(inputStr)}</pre></details>`;
               }
-              contentHtml += `<div class="streaming-block-tool-use"><div class="tool-use-header"><span class="tool-use-icon">&#9881;</span> <span class="tool-use-name">${this.escapeHtml(block.name || 'unknown')}</span></div>${inputHtml}</div>`;
+              const tn2 = block.name || 'unknown';
+              const foldable2 = tn2.startsWith('mcp__') || tn2 === 'Edit';
+              if (foldable2) {
+                contentHtml += `<details class="streaming-block-tool-use"><summary class="tool-use-header" style="cursor:pointer;user-select:none;list-style:none;"><span class="tool-use-icon">&#9881;</span> <span class="tool-use-name">${this.escapeHtml(tn2)}</span></summary>${inputHtml}</details>`;
+              } else {
+                contentHtml += `<div class="streaming-block-tool-use"><div class="tool-use-header"><span class="tool-use-icon">&#9881;</span> <span class="tool-use-name">${this.escapeHtml(tn2)}</span></div>${inputHtml}</div>`;
+              }
             } else if (block.type === 'tool_result') {
               const content = typeof block.content === 'string' ? block.content : JSON.stringify(block.content);
               const smartHtml = typeof StreamingRenderer !== 'undefined' ? StreamingRenderer.renderSmartContentHTML(content, this.escapeHtml.bind(this)) : `<pre class="tool-result-pre">${this.escapeHtml(content.length > 2000 ? content.substring(0, 2000) + '\n... (truncated)' : content)}</pre>`;

package/static/js/streaming-renderer.js

@@ -624,12 +624,31 @@ class StreamingRenderer {
    * Render tool use block with smart parameter display
    */
   renderBlockToolUse(block, context) {
-    const div = document.createElement('div');
-    div.className = 'block-tool-use';
-
     const toolName = block.name || 'unknown';
     const input = block.input || {};
+    const shouldFold = toolName.startsWith('mcp__') || toolName === 'Edit';
+
+    if (shouldFold) {
+      const details = document.createElement('details');
+      details.className = 'block-tool-use';
+      const summary = document.createElement('summary');
+      summary.className = 'tool-header';
+      summary.style.cssText = 'cursor:pointer;user-select:none;list-style:none;';
+      summary.innerHTML = `
+        <span class="tool-icon">${this.getToolIcon(toolName)}</span>
+        <span class="tool-name"><code>${this.escapeHtml(toolName)}</code></span>
+      `;
+      details.appendChild(summary);
+      if (Object.keys(input).length > 0) {
+        const paramsDiv = document.createElement('div');
+        paramsDiv.innerHTML = this.renderSmartParams(toolName, input);
+        details.appendChild(paramsDiv);
+      }
+      return details;
+    }
 
+    const div = document.createElement('div');
+    div.className = 'block-tool-use';
     div.innerHTML = `
       <div class="tool-header">
         <span class="tool-icon">${this.getToolIcon(toolName)}</span>
@@ -637,7 +656,6 @@ class StreamingRenderer {
       </div>
       ${Object.keys(input).length > 0 ? this.renderSmartParams(toolName, input) : ''}
     `;
-
     return div;
   }
 

package/static/js/voice.js

@@ -1,167 +1,24 @@
 (function() {
-  const BASE = window.__BASE_URL || '';
-  let STT = null;
-  let TTS = null;
+  var BASE = window.__BASE_URL || '';
+  var isRecording = false;
+  var ttsEnabled = true;
+  var voiceActive = false;
+  var lastSpokenBlockIndex = -1;
+  var currentConversationId = null;
+  var speechQueue = [];
+  var isSpeaking = false;
+  var currentAudio = null;
+  var mediaStream = null;
+  var audioContext = null;
+  var scriptNode = null;
+  var recordedChunks = [];
+  var TARGET_SAMPLE_RATE = 16000;
 
-  async function loadSDK() {
-    try {
-      const mod = await import(BASE + '/webtalk/sdk.js');
-      STT = mod.STT;
-      TTS = mod.TTS;
-      return true;
-    } catch (e) {
-      console.warn('Webtalk SDK load failed:', e.message);
-      return false;
-    }
-  }
-  let stt = null;
-  let tts = null;
-  let isRecording = false;
-  let ttsEnabled = true;
-  let voiceActive = false;
-  let lastSpokenBlockIndex = -1;
-  let currentConversationId = null;
-  let sttReady = false;
-  let ttsReady = false;
-  let speechQueue = [];
-  let isSpeaking = false;
-
-  async function init() {
+  function init() {
     setupTTSToggle();
     setupUI();
     setupStreamingListener();
     setupAgentSelector();
-    var sdkLoaded = await loadSDK();
-    if (sdkLoaded) {
-      initSTT();
-      initTTS();
-    } else {
-      sttLoadPhase = 'failed';
-      updateMicState();
-    }
-  }
-
-  var sttLoadPhase = 'starting';
-
-  async function initSTT() {
-    try {
-      stt = new STT({
-        basePath: BASE + '/webtalk',
-        onTranscript: function(text) {
-          var el = document.getElementById('voiceTranscript');
-          if (el) {
-            el.textContent = text;
-            el.setAttribute('data-final', text);
-          }
-        },
-        onPartial: function(text) {
-          var el = document.getElementById('voiceTranscript');
-          if (el) {
-            var existing = el.getAttribute('data-final') || '';
-            el.textContent = existing + text;
-          }
-        },
-        onStatus: function(status) {
-          var micBtn = document.getElementById('voiceMicBtn');
-          if (!micBtn) return;
-          if (status === 'recording') {
-            micBtn.classList.add('recording');
-          } else {
-            micBtn.classList.remove('recording');
-          }
-        }
-      });
-      var origInit = stt.init.bind(stt);
-      var initPromise = new Promise(function(resolve, reject) {
-        origInit().then(resolve).catch(reject);
-        if (stt.worker) {
-          var origHandler = stt.worker.onmessage;
-          stt.worker.onmessage = function(e) {
-            var msg = e.data;
-            if (msg && msg.status) {
-              if (msg.status === 'progress' || msg.status === 'download') {
-                if (sttLoadPhase !== 'downloading') {
-                  sttLoadPhase = 'downloading';
-                  updateMicState();
-                }
-              } else if (msg.status === 'done' && msg.file && msg.file.endsWith('.onnx')) {
-                sttLoadPhase = 'compiling';
-                updateMicState();
-              }
-            }
-            if (origHandler) origHandler.call(stt.worker, e);
-          };
-        }
-      });
-      await initPromise;
-      sttReady = true;
-      updateMicState();
-    } catch (e) {
-      console.warn('STT init failed:', e.message);
-      sttLoadPhase = 'failed';
-      updateMicState();
-    }
-  }
-
-  function updateMicState() {
-    var micBtn = document.getElementById('voiceMicBtn');
-    if (!micBtn) return;
-    if (sttReady) {
-      micBtn.removeAttribute('disabled');
-      micBtn.title = 'Click to record';
-      micBtn.classList.remove('loading');
-    } else if (sttLoadPhase === 'failed') {
-      micBtn.setAttribute('disabled', 'true');
-      micBtn.title = 'Speech recognition failed to load';
-      micBtn.classList.remove('loading');
-    } else {
-      micBtn.setAttribute('disabled', 'true');
-      micBtn.classList.add('loading');
-      if (sttLoadPhase === 'downloading') {
-        micBtn.title = 'Downloading speech models...';
-      } else if (sttLoadPhase === 'compiling') {
-        micBtn.title = 'Compiling speech models (may take a minute)...';
-      } else {
-        micBtn.title = 'Loading speech recognition...';
-      }
-    }
-  }
-
-  async function initTTS(retries) {
-    var maxRetries = retries || 3;
-    for (var attempt = 0; attempt < maxRetries; attempt++) {
-      try {
-        tts = new TTS({
-          basePath: BASE + '/webtalk',
-          apiBasePath: BASE,
-          onStatus: function() {},
-          onAudioReady: function(url) {
-            var audio = new Audio(url);
-            audio.onended = function() {
-              isSpeaking = false;
-              processQueue();
-            };
-            audio.onerror = function() {
-              isSpeaking = false;
-              processQueue();
-            };
-            audio.play().catch(function() {
-              isSpeaking = false;
-              processQueue();
-            });
-          }
-        });
-        await tts.init();
-        ttsReady = true;
-        return;
-      } catch (e) {
-        console.warn('TTS init attempt ' + (attempt + 1) + '/' + maxRetries + ' failed:', e.message);
-        tts = null;
-        if (attempt < maxRetries - 1) {
-          await new Promise(function(r) { setTimeout(r, 3000 * (attempt + 1)); });
-        }
-      }
-    }
   }
 
   function setupAgentSelector() {
@@ -203,6 +60,8 @@
   function setupUI() {
     var micBtn = document.getElementById('voiceMicBtn');
     if (micBtn) {
+      micBtn.removeAttribute('disabled');
+      micBtn.title = 'Click to record';
       micBtn.addEventListener('click', function(e) {
         e.preventDefault();
         if (!isRecording) {
@@ -216,43 +75,104 @@
     if (sendBtn) {
       sendBtn.addEventListener('click', sendVoiceMessage);
     }
-    updateMicState();
+  }
+
+  function resampleBuffer(inputBuffer, fromRate, toRate) {
+    if (fromRate === toRate) return inputBuffer;
+    var ratio = fromRate / toRate;
+    var newLen = Math.round(inputBuffer.length / ratio);
+    var result = new Float32Array(newLen);
+    for (var i = 0; i < newLen; i++) {
+      var srcIdx = i * ratio;
+      var lo = Math.floor(srcIdx);
+      var hi = Math.min(lo + 1, inputBuffer.length - 1);
+      var frac = srcIdx - lo;
+      result[i] = inputBuffer[lo] * (1 - frac) + inputBuffer[hi] * frac;
+    }
+    return result;
   }
 
   async function startRecording() {
     if (isRecording) return;
     var el = document.getElementById('voiceTranscript');
-    if (!stt || !sttReady) {
-      if (el) el.textContent = 'Speech recognition still loading, please wait...';
-      return;
-    }
     if (el) {
       el.textContent = '';
       el.setAttribute('data-final', '');
     }
-    isRecording = true;
     try {
-      await stt.startRecording();
+      mediaStream = await navigator.mediaDevices.getUserMedia({ audio: true });
+      audioContext = new (window.AudioContext || window.webkitAudioContext)();
+      var source = audioContext.createMediaStreamSource(mediaStream);
+      scriptNode = audioContext.createScriptProcessor(4096, 1, 1);
+      recordedChunks = [];
+      scriptNode.onaudioprocess = function(e) {
+        var data = e.inputBuffer.getChannelData(0);
+        recordedChunks.push(new Float32Array(data));
+      };
+      source.connect(scriptNode);
+      scriptNode.connect(audioContext.destination);
+      isRecording = true;
+      var micBtn = document.getElementById('voiceMicBtn');
+      if (micBtn) micBtn.classList.add('recording');
     } catch (e) {
       isRecording = false;
       if (el) el.textContent = 'Mic access denied or unavailable: ' + e.message;
-      console.warn('Recording start failed:', e.message);
     }
   }
 
   async function stopRecording() {
-    if (!stt || !isRecording) return;
+    if (!isRecording) return;
     isRecording = false;
+    var micBtn = document.getElementById('voiceMicBtn');
+    if (micBtn) micBtn.classList.remove('recording');
+    var el = document.getElementById('voiceTranscript');
+    if (scriptNode) { scriptNode.disconnect(); scriptNode = null; }
+    if (mediaStream) {
+      mediaStream.getTracks().forEach(function(t) { t.stop(); });
+      mediaStream = null;
+    }
+    var sourceSampleRate = audioContext ? audioContext.sampleRate : 48000;
+    if (audioContext) { audioContext.close().catch(function() {}); audioContext = null; }
+    if (recordedChunks.length === 0) return;
+    var totalLen = 0;
+    for (var i = 0; i < recordedChunks.length; i++) totalLen += recordedChunks[i].length;
+    var merged = new Float32Array(totalLen);
+    var offset = 0;
+    for (var j = 0; j < recordedChunks.length; j++) {
+      merged.set(recordedChunks[j], offset);
+      offset += recordedChunks[j].length;
+    }
+    recordedChunks = [];
+    var resampled = resampleBuffer(merged, sourceSampleRate, TARGET_SAMPLE_RATE);
+    if (el) el.textContent = 'Transcribing...';
     try {
-      await stt.stopRecording();
-    } catch (e) {}
+      var pcmBuffer = resampled.buffer;
+      var resp = await fetch(BASE + '/api/stt', {
+        method: 'POST',
+        headers: { 'Content-Type': 'application/octet-stream' },
+        body: pcmBuffer
+      });
+      var data = await resp.json();
+      if (data.text) {
+        if (el) {
+          el.textContent = data.text;
+          el.setAttribute('data-final', data.text);
+        }
+      } else if (data.error) {
+        if (el) el.textContent = 'Error: ' + data.error;
+      } else {
+        if (el) el.textContent = '';
+      }
+    } catch (e) {
+      if (el) el.textContent = 'Transcription failed: ' + e.message;
+    }
   }
 
   function sendVoiceMessage() {
     var el = document.getElementById('voiceTranscript');
     if (!el) return;
     var text = el.textContent.trim();
-    if (!text) return;
+    if (!text || text.startsWith('Transcribing') || text.startsWith('Error')) return;
     addVoiceBlock(text, true);
     el.textContent = '';
     el.setAttribute('data-final', '');
@@ -266,7 +186,7 @@
   }
 
   function speak(text) {
-    if (!ttsEnabled || !tts || !ttsReady) return;
+    if (!ttsEnabled) return;
     var clean = text.replace(/<[^>]*>/g, '').trim();
     if (!clean) return;
     speechQueue.push(clean);
@@ -277,7 +197,35 @@
     if (isSpeaking || speechQueue.length === 0) return;
     isSpeaking = true;
     var text = speechQueue.shift();
-    tts.generate(text).catch(function() {
+    fetch(BASE + '/api/tts', {
+      method: 'POST',
+      headers: { 'Content-Type': 'application/json' },
+      body: JSON.stringify({ text: text })
+    }).then(function(resp) {
+      if (!resp.ok) throw new Error('TTS failed');
+      return resp.blob();
+    }).then(function(blob) {
+      var url = URL.createObjectURL(blob);
+      currentAudio = new Audio(url);
+      currentAudio.onended = function() {
+        URL.revokeObjectURL(url);
+        currentAudio = null;
+        isSpeaking = false;
+        processQueue();
+      };
+      currentAudio.onerror = function() {
+        URL.revokeObjectURL(url);
+        currentAudio = null;
+        isSpeaking = false;
+        processQueue();
+      };
+      currentAudio.play().catch(function() {
+        URL.revokeObjectURL(url);
+        currentAudio = null;
+        isSpeaking = false;
+        processQueue();
+      });
+    }).catch(function() {
       isSpeaking = false;
       processQueue();
     });
@@ -286,7 +234,10 @@
   function stopSpeaking() {
     speechQueue = [];
     isSpeaking = false;
-    if (tts) tts.stop();
+    if (currentAudio) {
+      currentAudio.pause();
+      currentAudio = null;
+    }
   }
 
   function addVoiceBlock(text, isUser) {