remote-tools
Click on "Deploy Server".
Wait a few minutes for the server to deploy. Once ready, it will show a "Started" state.
In the chat, type
@followed by the MCP server name and your instructions, e.g., "@remote-toolsrun npm test on my local machine and report the results"
That's it! The server will respond to your query, and you can continue using it as needed.
Here is a step-by-step guide with screenshots.
remote-tools
让服务端 Claude Agent SDK agent 安全地操作用户本机:6 个 client__* MCP 工具经
WebSocket 隧道转发到用户桌面端(Plan B:remote-tools-desktop)执行。
设计文档:../docs/remote-tools/2026-09-09-remote-tool-proxy-design.md
宿主集成样例
import { query } from "@anthropic-ai/claude-agent-sdk";
import { TunnelServer, createServerWriteFence } from "remote-tools"; // 根入口不加载 SDK
import { createClientToolsSdkMcp } from "remote-tools/mcp-bridge/sdk";
const tunnel = new TunnelServer({ token: process.env.REMOTE_TOOLS_TOKEN! });
await tunnel.listen(8787);
const codingSession = query({
prompt: "按 docs/ 里的 Spec 与 Plan 编码",
options: {
mcpServers: { client: createClientToolsSdkMcp(tunnel) },
canUseTool: createServerWriteFence({ docsRoots: ["/data/docs"] }),
disallowedTools: ["Bash"], // 服务器不执行;构建调试走 client__bash
systemPrompt: {
type: "preset", preset: "claude_code",
append: "服务器存放需求与技术文档(docs/ 内可读可写);用户的所有代码、依赖与运行环境"
+ "都在其本机——编码、构建、调试、执行一律使用 client__* 工具。",
},
},
});导出位置:remote-tools 根入口是 SDK-free 的(TunnelServer / createToolHandlers /
createServerWriteFence);依赖 Agent SDK 的两个适配器分别走子路径
remote-tools/mcp-bridge/sdk(6 个 client__* 工具)与
remote-tools/phase/switch(submit_phase 阶段切换)。
阶段切换(submit_phase)与两阶段 options 的组合见设计文档 §10。
Related MCP server: rvn
重要注意事项
bash 默认超时 120s > SDK 默认 60s 流关闭超时:宿主需设置环境变量
CLAUDE_CODE_STREAM_CLOSE_TIMEOUT(如180000),否则长构建/测试会被 SDK 提前切断。隧道令牌用 32 字节随机值生成:
node -e "console.log(require('crypto').randomBytes(32).toString('hex'))"。客户端桌面端(Tauri)是独立项目 remote-tools-desktop,协议为
src/protocol/envelope.ts定义的 JSON 信封(TS 与 Rust 仅结构对齐)。集成样例已按当前安装的 Agent SDK 用
tsc校验:该版本的Options没有appendSystemPrompt字段,追加指令走systemPrompt: { type: "preset", preset: "claude_code", append }; 若你的 SDK 版本提供appendSystemPrompt,两者语义等价。
开发
npm test # vitest
npm run typecheckThis server cannot be deployed
Maintenance
Related MCP Connectors
Securely control computers you explicitly pair through files, terminals, processes, screenshots, desktop UI/input, clipboard, browser automation, diagnostics, and document tools.
Operate Linux, macOS and Windows from your LLM. Every action runs through an auditable allowlist.
Your org's AI agents, tasks, runs, search, and brain files as MCP tools and resources.
Remote MCP server to read and manage your Atako AI agents, messages, files, and integrations.
Related MCP Servers
- FlicenseNot gradedqualityAmaintenanceEnables MCP clients to safely access user-local filesystems, apply validated patches, inspect git state, and run persistent jobs on outbound-connected local runners through a stateless Cloudflare control plane.8-
- AlicenseNot gradedqualityAmaintenanceEnables AI clients to securely control and interact with a local Windows machine through 218 configurable tools for files, Git, processes, Windows UI, browser automation, WSL, Office, recovery, skills, and child MCP servers.5 npmMIT
- FlicenseNot gradedqualityCmaintenanceEnables agents to connect to remote MCP servers once, access their tools through a compact MCP endpoint, pair a CLI inside sandboxes, and create watches that turn command or tool output into pollable structured events.2-
- AlicenseNot gradedqualityBmaintenanceEnables agent clients to safely connect to tools and execution resources through MCP with authorization, approvals, audit, chat-context isolation, SSH/Docker access, and long-running command session tracking.MIT