openclaw-speech-input 2026.3.13

package/README.md

@@ -0,0 +1,166 @@
+# 🎤 Voice Input Plugin
+
+语音输入插件 —— 通过浏览器麦克风进行语音识别，直接与 OpenClaw 对话。
+
+![Version](https://img.shields.io/badge/version-1.0.0-blue)
+![Platform](https://img.shields.io/badge/platform-Windows%20%7C%20Linux%20%7C%20Mac-orange)
+![Browser](https://img.shields.io/badge/browser-Chrome%20%7C%20Edge-green)
+
+## ✨ 功能特性
+
+| 功能 | 说明 |
+|------|------|
+| 🎤 语音识别 | 浏览器原生 Web Speech API，实时语音转文字 |
+| 💬 文字输入 | 支持手动输入文字进行对话 |
+| 🔊 语音合成 | TTS 语音播报 AI 回复 |
+| 🔑 自动认证 | 自动注入 Gateway Token，无需手动配置 |
+| 🌐 多语言支持 | 支持中文、英文等多种语言识别 |
+
+## 📁 文件结构
+
+```
+openclaw-speech-input/
+├── index.ts              # 插件入口 (TypeScript)
+├── web.html              # 前端交互页面
+├── openclaw.plugin.json  # 插件配置文件
+└── README.md             # 使用文档
+```
+
+## 🚀 安装方法
+
+### 方式一：解压安装（推荐）
+
+1. 解压插件包文件
+2. 将 `openclaw-speech-input` 文件夹复制到插件目录：
+   - **Windows:** `C:\Users\<用户名>\.openclaw\extensions\`
+   - **Linux/Mac:** `~/.openclaw/extensions/`
+   - **或使用环境变量:** `$env:OPENCLAW_HOME\.openclaw\extensions\`
+3. 重启 Gateway：
+   ```bash
+   openclaw gateway restart
+   ```
+
+### 方式二：手动安装
+
+1. 创建插件目录：
+   ```bash
+   # Windows
+   mkdir "C:\Users\<用户名>\.openclaw\extensions\openclaw-speech-input"
+   
+   # Linux/Mac
+   mkdir -p ~/.openclaw/extensions/openclaw-speech-input
+   ```
+2. 复制以下文件到该目录：
+   - `index.ts`
+   - `web.html`
+   - `openclaw.plugin.json`
+3. 重启 Gateway：
+   ```bash
+   openclaw gateway restart
+   ```
+
+## 📖 使用方法
+
+### 方式一：命令打开（推荐）
+
+在聊天中输入命令：
+```
+/stt
+```
+
+或直接说**"打开语音输入"**，插件会自动识别并打开。
+
+### 方式二：手动打开
+
+在浏览器中访问：
+```
+http://127.0.0.1:18789/plugin/openclaw-speech-input/
+```
+
+### 使用流程
+
+1. **授权麦克风** —— 首次使用需要允许浏览器访问麦克风
+2. **点击麦克风按钮** 🎤 —— 开始语音输入
+3. **说话** —— 语音会自动识别为文字
+4. **发送** —— 再次点击按钮或按 Enter 发送
+5. **接收回复** —— AI 回复会以文字显示
+
+> **💡 提示：** 插件会自动从 OpenClaw 配置中获取 Gateway Token，无需手动配置。
+
+---
+
+## ⚙️ 配置说明
+
+| 配置项 | 说明 | 默认值 |
+|--------|------|--------|
+| Gateway 地址 | HTTP 服务地址 | `http://127.127.0.1:18789` |
+| 识别语言 | 语音识别语言 | 中文（普通话） |
+| TTS 语音 | 语音合成声音 | 浏览器默认 |
+
+---
+
+## ❓ 常见问题
+
+### 1. 语音识别不工作
+
+- **浏览器不支持** —— 推荐使用 Chrome、Edge 或 Safari
+- **未授权麦克风** —— 检查浏览器地址栏左侧的麦克风图标，确保已授权
+
+### 2. 无法连接 Gateway
+
+- 确认 Gateway 正在运行：`openclaw gateway status`
+- 检查端口 18789 是否被占用：`netstat -ano | findstr 18789`
+
+### 3. 麦克风权限被拒绝
+
+- 在浏览器设置中允许麦克风访问
+- Chrome/Edge 用户可访问 `chrome://settings/content/microphone` 进行管理
+
+### 4. 401 未授权错误
+
+- 重启 Gateway：`openclaw gateway restart`
+- 检查 `gateway.auth.token` 配置是否正确
+
+### 5. 语音识别响应慢
+
+- 检查网络连接状况
+- 尝试更换识别语言设置
+- 关闭其他占用麦克风的应用程序
+
+---
+
+## 🗑️ 卸载插件
+
+1. 删除插件目录：
+   ```bash
+   # Windows
+   rmdir /s /q "C:\Users\<用户名>\.openclaw\extensions\openclaw-speech-input"
+   
+   # Linux/Mac
+   rm -rf ~/.openclaw/extensions/openclaw-speech-input
+   ```
+2. 重启 Gateway：
+   ```bash
+   openclaw gateway restart
+   ```
+
+---
+
+## 🔧 技术实现
+
+| 技术 | 说明 |
+|------|------|
+| 语音识别 | [Web Speech API](https://developer.mozilla.org/en-US/docs/Web/API/SpeechRecognition) |
+| 语音合成 | [SpeechSynthesis API](https://developer.mozilla.org/en-US/docs/Web/API/SpeechSynthesis) |
+| 通信协议 | OpenAI 兼容 HTTP API |
+| 认证方式 | Bearer Token（自动从配置注入） |
+
+---
+
+## 📝 更新日志
+
+### v2026.3.13 (2026-03-22)
+- 🎉 初始版本发布
+- 支持语音识别
+- 自动认证功能
+- 指令打开语音输入

package/index.ts

@@ -0,0 +1,206 @@
+import type { OpenClawPluginApi } from "openclaw/plugin-sdk";
+import { emptyPluginConfigSchema, runPluginCommandWithTimeout } from "openclaw/plugin-sdk";
+import { readFileSync } from "node:fs";
+import { join } from "node:path";
+
+// ─── Constants ───────────────────────────────────────────────────────────────
+
+const PLUGIN_ID = "openclaw-speech-input";
+const ROUTE_PATH = "/plugin/openclaw-speech-input/";
+const CHAT_URI = "/v1/chat/completions";
+const COMMAND_TIMEOUT_MS = 5000;
+
+// ─── Platform helpers ─────────────────────────────────────────────────────────
+
+/**
+ * Returns the platform-appropriate command to open a URL in the default browser.
+ */
+function getOpenBrowserCommand(url: string): string[] {
+  switch (process.platform) {
+    case "win32":
+      // Use cmd /c start to avoid PowerShell execution-policy issues
+      return ["cmd", "/c", "start", "", url];
+    case "darwin":
+      return ["open", url];
+    default:
+      return ["xdg-open", url];
+  }
+}
+
+function getGatewayFullUrl(api: OpenClawPluginApi): string {
+  const gatewayCfg = api.config.gateway;
+
+  // 1. 若配置中显式定义了 remote url，以远程 URL 为绝对最高优先级
+  if (gatewayCfg?.mode === "remote" && gatewayCfg?.remote?.url) {
+    // 剔除末尾可能存在的斜杠，保证统一
+    return gatewayCfg.remote.url.replace(/\/$/, "");
+  }
+
+  // 2. 本地模式 fallback 拼接
+  // 使用 ?? 运算符确保默认值的有效兜底
+  const protocol = gatewayCfg?.tls?.enabled ? "https" : "http";
+
+  // 兼容 bind 为 'loopback' (默认) 或 IP 地址的情况
+  const rawBind = gatewayCfg?.bind || "127.0.0.1";
+  const host = rawBind === "loopback" ? "127.0.0.1" : rawBind;
+
+  const port = gatewayCfg?.port || 18789; // Openclaw 默认本地网关端口
+
+  return `${protocol}://${host}:${port}`;
+}
+
+// ─── Plugin definition ────────────────────────────────────────────────────────
+
+const voiceInputPlugin = {
+  id: PLUGIN_ID,
+  name: PLUGIN_ID,
+  description: "语音输入插件 — 通过浏览器麦克风进行语音识别并与 OpenClaw 对话",
+  configSchema: emptyPluginConfigSchema(),
+
+  register(api: OpenClawPluginApi): void {
+    // Read the HTML once at registration time (not per-request) for performance.
+    // The file is bundled alongside index.ts, so __dirname is reliable.
+    let htmlTemplate: string;
+    try {
+      htmlTemplate = readFileSync(join(__dirname, "web.html"), "utf-8");
+    } catch (err) {
+      api.logger.error(`[${api.name}] Failed to read web.html: ${String(err)}`);
+      // Register a stub so the plugin doesn't hard-crash the host process
+      htmlTemplate = `<!doctype html><html><body><p>Plugin asset missing: web.html</p></body></html>`;
+    }
+
+    // ── Command: stt ──────────────────────────────────────────────────────────
+    api.registerCommand({
+      name: "stt",
+      description: "打开语音输入页面",
+      requireAuth: false,
+
+      handler: async (ctx) => {
+        const webUrl = `${getGatewayFullUrl(api)}${ROUTE_PATH}`;
+
+        api.logger.info(`[${api.name}] Opening voice-input page: ${webUrl}`);
+
+        try {
+          const result = await runPluginCommandWithTimeout({
+            argv: getOpenBrowserCommand(webUrl),
+            timeoutMs: COMMAND_TIMEOUT_MS,
+          });
+          api.logger.info(`[${api.name}] Browser open result: ${JSON.stringify(result)}`);
+        } catch (err) {
+          const message = err instanceof Error ? err.message : String(err);
+          api.logger.warn(`[${api.name}] Could not auto-open browser: ${message}`);
+        }
+
+        return {
+          text: `正在打开语音输入界面，如果长时间未响应，您也可以手动[打开语音输入界面](${ROUTE_PATH})。`,
+        };
+      },
+    });
+
+    // ── HTTP route: serve the voice-input UI ──────────────────────────────────
+    api.registerHttpRoute({
+      path: ROUTE_PATH,
+      auth: "plugin",
+      match: "exact",
+
+      handler: async (_, res) => {
+        res.writeHead(200, {
+          "Content-Type": "text/html; charset=utf-8",
+          // Allow microphone/camera access from the plugin's own origin only
+          "Permissions-Policy": "microphone=(self), camera=(self)",
+          // Basic clickjacking protection
+          "X-Frame-Options": "SAMEORIGIN",
+          // Prevent MIME-type sniffing
+          "X-Content-Type-Options": "nosniff",
+          // Cache-Control: no-store so the injected token is never cached
+          "Cache-Control": "no-store",
+        });
+        res.end(htmlTemplate);
+
+        return true;
+      },
+    });
+
+    // ── HTTP route: proxy chat completions (SSE) ─────────────────────────────
+    api.registerHttpRoute({
+      path: `${ROUTE_PATH}chat`,
+      auth: "plugin",
+      match: "exact",
+
+      handler: async (req, res) => {
+        if (req.method !== "POST") {
+          res.writeHead(405).end("Method Not Allowed");
+          return true;
+        }
+
+        let bodyStr = "";
+        for await (const chunk of req) {
+          bodyStr += chunk.toString();
+        }
+
+        let clientBody: any;
+        try {
+          clientBody = JSON.parse(bodyStr);
+        } catch {
+          res.writeHead(400).end("Invalid JSON");
+          return true;
+        }
+
+        const token = (api.config.gateway?.auth?.token as string | undefined) ?? "";
+        const targetUrl = `http://${req.headers.host}${CHAT_URI}`;
+
+        // 构建转发请求头，所有鉴权和路由参数均由服务端配置，不依赖前端传递
+        const headers: Record<string, string> = {
+          "Content-Type": "application/json",
+        };
+        if (token) {
+          headers["Authorization"] = `Bearer ${token}`;
+        }
+
+        // 固定注入会话标识和 Agent ID，无需前端传递
+        headers["x-openclaw-session-key"] = "agent:main:openclaw-speech-input";
+        headers["x-openclaw-agent-id"] = "main";
+
+        try {
+          // 仅转发消息体内容，不修改结构
+          const upstreamRes = await fetch(targetUrl, {
+            method: "POST",
+            headers,
+            body: JSON.stringify(clientBody),
+          });
+
+          // 透传状态码和内容类型
+          res.writeHead(upstreamRes.status, {
+            "Content-Type": upstreamRes.headers.get("content-type") || "text/event-stream",
+            "Cache-Control": "no-cache",
+            Connection: "keep-alive",
+          });
+
+          if (!upstreamRes.body) {
+            res.end();
+            return true;
+          }
+
+          // 管道式流转发
+          for await (const chunk of upstreamRes.body) {
+            if (res.destroyed) break;
+            res.write(chunk);
+          }
+          res.end();
+        } catch (err) {
+          const msg = err instanceof Error ? err.message : "Proxy error";
+          if (!res.headersSent) {
+            res.writeHead(502, { "Content-Type": "application/json" });
+            res.end(JSON.stringify({ error: msg }));
+          } else {
+            res.destroy();
+          }
+        }
+
+        return true;
+      },
+    });
+  },
+};
+
+export default voiceInputPlugin;

package/openclaw.plugin.json

@@ -0,0 +1,15 @@
+{
+  "id": "openclaw-speech-input",
+  "name": "Openclaw Speech Input",
+  "description": "语音输入插件 - 通过浏览器麦克风进行语音识别并与 OpenClaw 对话",
+  "version": "2026.3.13",
+  "author": "Haoyue",
+  "configSchema": {
+    "type": "object",
+    "additionalProperties": false,
+    "properties": {}
+  },
+  "openclaw": {
+    "extensions": ["./index.ts"]
+  }
+}

package/package.json

@@ -0,0 +1,37 @@
+{
+  "name": "openclaw-speech-input",
+  "version": "2026.3.13",
+  "description": "语音输入插件 - 通过浏览器麦克风进行语音识别并与 OpenClaw 对话",
+  "keywords": [
+    "openclaw",
+    "speech"
+  ],
+  "license": "MIT",
+  "author": "Haoyue",
+  "files": [
+    "index.ts",
+    "web.html",
+    "openclaw.plugin.json",
+    "package.json",
+    "tsconfig.json"
+  ],
+  "type": "module",
+  "main": "index.ts",
+  "publishConfig": {
+    "access": "public",
+    "registry": "https://registry.npmjs.org/"
+  },
+  "devDependencies": {
+    "@types/node": "^25.2.0",
+    "typescript": "^5.3.0"
+  },
+  "peerDependencies": {
+    "openclaw": ">=2026.3.11"
+  },
+  "openclaw": {
+    "extensions": [
+      "./index.ts"
+    ],
+    "installDependencies": true
+  }
+}

package/tsconfig.json

@@ -0,0 +1,16 @@
+{
+  "compilerOptions": {
+    "target": "ES2022",
+    "module": "NodeNext",
+    "moduleResolution": "NodeNext",
+    "declaration": true,
+    "outDir": "./dist",
+    "rootDir": ".",
+    "strict": true,
+    "esModuleInterop": true,
+    "skipLibCheck": true,
+    "resolveJsonModule": true
+  },
+  "include": ["index.ts", "src/**/*.ts"],
+  "exclude": ["node_modules", "dist"]
+}