local-assistant-mcp
Allows AI assistants like ChatGPT to securely connect to local development tools (file operations, commands, browser, and desktop control) via an OpenAI Platform tunnel.
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., "@local-assistant-mcpopen the browser and search for 'MCP servers'"
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.
本地开发助手(Local MCP Server)
Windows 本地开发 MCP Server,让 AI 在明确的工作区边界内调用文件、命令、浏览器和桌面工具。
核心能力
工作区隔离:先绑定项目目录,再通过独立
workspace_handle操作,避免不同项目互相影响。文件与代码操作:读取、搜索、创建、修改和删除项目文件,并对正式写入提供 Git / 备份保护。
命令与进程:运行受控开发命令、测试和后台进程,并限制工作目录和执行边界。
浏览器工具:支持页面结构读取、网络检查和隔离浏览器验证。
桌面控制:支持窗口、鼠标、键盘和截图等本地桌面操作。
临时文件隔离:诊断脚本和一次性文件放在项目目录之外,避免污染正式代码。
Secure MCP Tunnel:让 ChatGPT 通过受控通道连接本机 MCP;本地 MCP 仍只监听本机地址,不直接开放公网端口。
Related MCP server: DeskTether MCP Server
运行环境
Windows 10 / 11 64 位
Python 3.12 x64
Windows PowerShell 5.1 或更高版本
首次安装需要联网下载 Python 依赖和 Playwright Chromium
安装与启动
在仓库目录打开 PowerShell:
.\install.ps1
.\doctor.ps1
.\start.ps1默认地址:
MCP: http://127.0.0.1:18766/mcp
Health: http://127.0.0.1:18766/health停止服务:
.\stop.ps1如果本机端口冲突,可以指定其他端口:
.\start.ps1 -Port 18776
.\doctor.ps1 -Port 18776
.\stop.ps1 -Port 18776如果 PowerShell 阻止本地脚本执行:
powershell.exe -NoProfile -ExecutionPolicy Bypass -File .\install.ps1连接 ChatGPT
本地 MCP 可以独立运行;只有 ChatGPT 需要访问本机 MCP 时才需要 Tunnel。
先在 OpenAI Platform 创建 Tunnel,并准备对应的 Runtime API key,然后执行:
.\setup-tunnel.ps1 -TunnelId tunnel_...脚本会下载官方 openai/tunnel-client、校验 SHA256,并通过隐藏输入保存 Runtime API key。
启动 Tunnel:
.\start-tunnel.ps1
.\status-tunnel.ps1
.\doctor.ps1 -Tunnel停止 Tunnel:
.\stop-tunnel.ps1Tunnel 配置、客户端和加密后的 Runtime API key 默认保存在 %LOCALAPPDATA%\LocalAssistantMCP,不会写入仓库。
安全边界
MCP 只监听
127.0.0.1,不会直接绑定公网地址。项目操作受工作区边界约束,跨工作区操作会被拒绝。
沙箱执行和受控主机命令分别处理不同权限需求,主机命令仍受路径和参数限制。
Runtime API key 使用 Windows CurrentUser DPAPI 加密保存。
带账号密码的代理地址不会写入配置文件。
桌面控制可以操作真实窗口,只应连接你信任的 MCP 客户端。
测试
安装开发依赖并运行完整测试:
.\install.ps1 -Dev
.\run_tests.ps1License
MIT License。见 LICENSE。
This server cannot be deployed
Maintenance
Related MCP Connectors
Local Commander connects ChatGPT, Claude, Cursor and other MCP clients to Windows 11 computers you own. Your AI can read approved project files and Git state, run the checks and commands you authorize, and check local services. Your workspace at localcmd.app shows which computer ran each request and how it ended. Once you approve a computer and connect your AI, it can do what your Windows account can do; you can limit it to folders, block commands or switch it off at any time. Free on up to 5 computers; the Team pack adds 10 more for $15 a month.
Connect AI assistants to authorized Windows and macOS devices for files, commands, and management.
11Persistent cloud workspaces for AI agents: run commands, edit files, use git and a browser.
Use your Mac, Windows or Linux computer from ChatGPT, Claude or Codex: files, commands, documents.
Related MCP Servers
- FlicenseNot gradedqualityBmaintenanceEnables secure remote development on Windows by allowing LLM clients to manage files, run commands, and automate browsers within sandboxed workspaces, with multi-project switching and OAuth authentication.-
- AlicenseNot gradedqualityBmaintenanceEnables AI agents to safely perform file operations, run terminal sessions, manage processes, search content, and inspect Git repositories on a local Windows machine under configurable permission and audit controls.Apache 2.0
- AlicenseNot gradedqualityAmaintenanceEnables AI assistants to operate and monitor Windows workspaces through native tools for filesystem operations, PowerShell execution, background tasks, Windows services, GPU telemetry, PDF generation, and a built-in web gateway/dashboard.139 npm2MIT
- AlicenseNot gradedqualityBmaintenanceEnables AI agents to safely work on Windows developer workstations through governed, project-scoped filesystem access, durable command execution, bounded search, Windows operations, and resource-aware runtime controls. It also supports optional developer-context backends for code navigation, repository graph exploration, and memory retrieval.Apache 2.0