@foxden-app/foxclaw 0.5.48 → 0.5.50

package/.env.example

@@ -56,17 +56,25 @@ CODEX_CLI_BIN=/absolute/path/to/codex
 # Optional Telegram voice summaries.
 # When enabled, completed final answers get a "listen" button and /voice can
 # generate Telegram voice messages. SSH mode runs TTS and ffmpeg conversion on
-# the speech host; HTTP mode calls VOICE_TTS_URL and converts locally.
+# your own speech host; HTTP mode calls your own VOICE_TTS_URL and converts
+# locally. FoxClaw does not ship with a public TTS backend.
 # VOICE_TTS_ENABLED=false
+# HTTP backend example: compatible service exposing /v1/tts/custom or
+# /v1/tts/design and returning WAV audio. Local ffmpeg is required.
+# VOICE_TTS_MODE=http
+# VOICE_TTS_URL=http://127.0.0.1:18081
+# VOICE_TTS_TOKEN=
+# VOICE_FFMPEG_BIN=ffmpeg
+# SSH backend example: remote host has the compatible service, token in its
+# service .env, and ffmpeg available on the remote host.
 # VOICE_TTS_MODE=ssh
-# VOICE_TTS_SSH_HOST=thinkbook16p
-# VOICE_TTS_SSH_DIR=/home/wuya/dev/qwen-speech-server
+# VOICE_TTS_SSH_HOST=<ssh-host>
+# VOICE_TTS_SSH_DIR=/path/to/qwen-speech-server
 # VOICE_TTS_DESIGN_INSTRUCT=用自然清晰的中文女声朗读，语速适中。
 # VOICE_SUMMARY_BUTTON_ENABLED=true
+# VOICE_SUMMARY_TEXT_LIMIT=180
 # VOICE_TEXT_LIMIT=2800
-# VOICE_TTS_URL=https://tts.foxden.app
-# VOICE_TTS_TOKEN=
-# VOICE_FFMPEG_BIN=ffmpeg
+# VOICE_TTS_TIMEOUT_MS=120000
 
 # Optional: standard HTTP(S) proxy for Telegram and ChatGPT/Codex backend requests.
 # Put these in the same env file that `foxclaw start` installs into systemd/launchd.

package/CHANGELOG.md

@@ -2,6 +2,30 @@
 
 All notable FoxClaw changes are listed here. Each release note is bilingual so GitHub Releases and the npm package are useful to both Chinese and English readers.
 
+## 0.5.50 - 2026-06-20
+
+### 中文
+- 给 Telegram “听总结”按钮增加独立的 `VOICE_SUMMARY_TEXT_LIMIT`，默认只朗读短摘要，避免把完整最终回复交给慢速 TTS 后端导致长时间排队或超时。
+- 新增 `VOICE_TTS_TIMEOUT_MS`，HTTP/SSH 语音后端共用该超时配置；SSH 超时会终止远端调用，HTTP 超时会中止请求。
+- `/voice <文本>` 仍使用 `VOICE_TEXT_LIMIT`，方便手动朗读较长文本；最终总结按钮走更短的摘要上限，优先保证可听、及时返回。
+
+### English
+- Added a separate `VOICE_SUMMARY_TEXT_LIMIT` for Telegram "Listen" buttons, so final-answer voice summaries read a short digest instead of sending the entire final response to slow TTS backends.
+- Added `VOICE_TTS_TIMEOUT_MS` for both HTTP and SSH voice backends; SSH calls are terminated on timeout and HTTP requests are aborted.
+- `/voice <text>` still uses `VOICE_TEXT_LIMIT` for manual longer reads, while final-answer buttons use the shorter summary limit for timely playback.
+
+## 0.5.49 - 2026-06-20
+
+### 中文
+- 将 Telegram 总结语音后端彻底改为用户自配：默认不再内置 `thinkbook16p`、`tts.foxden.app` 或任何私有 TTS 服务地址。
+- 启用语音后如果缺少 `VOICE_TTS_URL` 或 `VOICE_TTS_SSH_HOST`/`VOICE_TTS_SSH_DIR`，会明确提示配置缺失，而不是尝试连接开发者私有服务。
+- 更新 `.env.example`，把 HTTP/SSH 语音后端写成通用占位示例，明确 FoxClaw 不随包提供公共 TTS 后端。
+
+### English
+- Made the Telegram voice-summary backend fully user-configured: FoxClaw no longer embeds `thinkbook16p`, `tts.foxden.app`, or any private TTS service endpoint as a product default.
+- When voice is enabled without `VOICE_TTS_URL` or `VOICE_TTS_SSH_HOST`/`VOICE_TTS_SSH_DIR`, FoxClaw now reports the missing configuration instead of trying a developer-private service.
+- Updated `.env.example` to show generic HTTP/SSH backend placeholders and state that FoxClaw does not ship with a public TTS backend.
+
 ## 0.5.48 - 2026-06-20
 
 ### 中文

package/dist/config.d.ts

@@ -67,12 +67,14 @@ export interface AppConfig {
     voiceTtsMode: 'http' | 'ssh';
     voiceTtsUrl: string | null;
     voiceTtsToken: string | null;
-    voiceTtsSshHost: string;
-    voiceTtsSshDir: string;
+    voiceTtsSshHost: string | null;
+    voiceTtsSshDir: string | null;
     voiceTtsDesignInstruct: string;
     voiceFfmpegBin: string;
     voiceSummaryButtonEnabled: boolean;
+    voiceSummaryTextLimit: number;
     voiceTextLimit: number;
+    voiceTtsTimeoutMs: number;
 }
 export declare function loadConfig(): AppConfig;
 export declare function selectDefaultRuntimeBotToken(configuredTokens: string[], legacyToken: string | null): string | null;

package/dist/config.js

@@ -98,15 +98,17 @@ export function loadConfig() {
         authSyncTempDir: process.env.AUTH_SYNC_TEMP_DIR || DEFAULT_AUTH_SYNC_TEMP_DIR,
         authAutoDeleteNeedsRepair: boolEnv('AUTH_AUTO_DELETE_NEEDS_REPAIR', false),
         voiceTtsEnabled: boolEnv('VOICE_TTS_ENABLED', false),
-        voiceTtsMode: parseVoiceTtsMode(process.env.VOICE_TTS_MODE || (process.env.VOICE_TTS_URL?.trim() ? 'http' : 'ssh')),
-        voiceTtsUrl: optional('VOICE_TTS_URL') ?? 'https://tts.foxden.app',
+        voiceTtsMode: parseVoiceTtsMode(process.env.VOICE_TTS_MODE || (process.env.VOICE_TTS_SSH_HOST?.trim() ? 'ssh' : 'http')),
+        voiceTtsUrl: optional('VOICE_TTS_URL'),
         voiceTtsToken: optional('VOICE_TTS_TOKEN'),
-        voiceTtsSshHost: process.env.VOICE_TTS_SSH_HOST?.trim() || 'thinkbook16p',
-        voiceTtsSshDir: process.env.VOICE_TTS_SSH_DIR?.trim() || '/home/wuya/dev/qwen-speech-server',
+        voiceTtsSshHost: optional('VOICE_TTS_SSH_HOST'),
+        voiceTtsSshDir: optional('VOICE_TTS_SSH_DIR'),
         voiceTtsDesignInstruct: process.env.VOICE_TTS_DESIGN_INSTRUCT?.trim() || '用自然清晰的中文女声朗读，语速适中。',
         voiceFfmpegBin: process.env.VOICE_FFMPEG_BIN?.trim() || 'ffmpeg',
         voiceSummaryButtonEnabled: boolEnv('VOICE_SUMMARY_BUTTON_ENABLED', true),
+        voiceSummaryTextLimit: intEnv('VOICE_SUMMARY_TEXT_LIMIT', 180),
         voiceTextLimit: intEnv('VOICE_TEXT_LIMIT', 2800),
+        voiceTtsTimeoutMs: intEnv('VOICE_TTS_TIMEOUT_MS', 120_000),
     };
     ensureAppDirs(config);
     return config;

package/dist/controller/controller.js

@@ -8394,7 +8394,7 @@ export class BridgeSessionCore {
     }
     registerVoiceSnippet(scopeId, text) {
         this.pruneVoiceSnippets();
-        const normalized = normalizeVoiceText(text, this.config.voiceTextLimit);
+        const normalized = normalizeVoiceText(text, this.config.voiceSummaryTextLimit);
         const id = crypto.randomBytes(6).toString('hex');
         this.voiceSnippets.set(id, { scopeId, text: normalized, createdAt: Date.now() });
         this.latestVoiceSnippetByScope.set(scopeId, id);

package/dist/voice/tts.js

@@ -61,14 +61,17 @@ async function requestTtsWav(config, pathname, body) {
         throw new Error('VOICE_TTS_URL is not configured');
     }
     const endpoint = new URL(pathname, config.voiceTtsUrl.endsWith('/') ? config.voiceTtsUrl : `${config.voiceTtsUrl}/`);
+    const abortController = new AbortController();
+    const timer = setTimeout(() => abortController.abort(), config.voiceTtsTimeoutMs);
     const response = await fetch(endpoint, {
         method: 'POST',
+        signal: abortController.signal,
         headers: {
             'content-type': 'application/json',
             ...(config.voiceTtsToken ? { authorization: `Bearer ${config.voiceTtsToken}` } : {}),
         },
         body: JSON.stringify(body),
-    });
+    }).finally(() => clearTimeout(timer));
     if (!response.ok) {
         throw new Error(`TTS request failed: ${response.status} ${response.statusText}: ${await response.text().catch(() => '')}`);
     }
@@ -105,6 +108,12 @@ async function convertToOggOpus(input, ffmpegBin) {
     }
 }
 async function synthesizeViaSsh(text, config) {
+    if (!config.voiceTtsSshHost) {
+        throw new Error('VOICE_TTS_SSH_HOST is not configured');
+    }
+    if (!config.voiceTtsSshDir) {
+        throw new Error('VOICE_TTS_SSH_DIR is not configured');
+    }
     const encodedText = Buffer.from(text, 'utf8').toString('base64');
     const script = String.raw `set -euo pipefail
 TEXT="$(printf '%s' "$1" | base64 -d)"
@@ -166,9 +175,9 @@ PY
         encodedText,
         config.voiceTtsSshDir,
         config.voiceTtsDesignInstruct,
-    ], script);
+    ], script, config.voiceTtsTimeoutMs);
 }
-async function runSshBinary(args, stdin) {
+async function runSshBinary(args, stdin, timeoutMs) {
     return new Promise((resolve, reject) => {
         const child = spawn('ssh', args, { stdio: ['pipe', 'pipe', 'pipe'] });
         const stdout = [];
@@ -176,7 +185,7 @@ async function runSshBinary(args, stdin) {
         const timer = setTimeout(() => {
             child.kill('SIGTERM');
             reject(new Error('remote TTS timed out'));
-        }, 120_000);
+        }, timeoutMs);
         child.stdout.on('data', (chunk) => {
             stdout.push(Buffer.isBuffer(chunk) ? chunk : Buffer.from(chunk));
         });

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@foxden-app/foxclaw",
-  "version": "0.5.48",
+  "version": "0.5.50",
   "description": "Foxden local execution claw for controlling Codex from trusted chat interfaces.",
   "type": "module",
   "main": "dist/main.js",