open-ssh-mcp
An MCP server for managing remote SSH servers, enabling AI agents to execute commands, transfer files, and perform deployment operations securely.
README
SSH MCP Server
基于 Model Context Protocol (MCP) 的 SSH 远程服务器管理工具,为 AI 代理提供标准化的 SSH 运维能力。使用 FastMCP + asyncssh 构建。
特性
- 连接管理 — 懒连接、健康检查、自动重连、空闲超时
- 命令执行 — 远程命令、sudo(策略门控)、批量执行
- 文件操作 — SFTP 上传 / 下载 / 读写 / 删除 / 列表 / 元数据
- 部署运维 — 完整部署流程:上传 → 备份 → 放置 → 重启
- 安全设计 — 命令黑名单、路径限制、审计日志、确认门控、主机密钥验证
- 多服务器 — 单一配置文件管理多台服务器
快速开始
1. 安装
方式一:下载预编译二进制(推荐)
前往 Releases 下载对应平台的可执行文件:
| 平台 | 文件 |
|---|---|
| macOS Apple Silicon | open-ssh-mcp-darwin-arm64 |
| macOS Intel | open-ssh-mcp-darwin-amd64 |
| Linux x86_64 | open-ssh-mcp-linux-amd64 |
| Linux ARM64 | open-ssh-mcp-linux-arm64 |
| Windows x64 | open-ssh-mcp-windows-amd64.exe |
下载后赋予执行权限:
chmod +x open-ssh-mcp-darwin-arm64
方式二:从源码安装
pip install -e .
2. 配置
复制示例配置并编辑:
cp config.example.json config.json
完整配置说明见 配置说明。
3. 在 Qoder 中配置
打开 Qoder → Settings → MCP Servers,添加以下配置:
使用预编译二进制(推荐):
{
"mcpServers": {
"ssh": {
"command": "/path/to/open-ssh-mcp-darwin-arm64",
"args": ["--config", "/path/to/config.json"]
}
}
}
使用 uv 运行(开发模式):
{
"mcpServers": {
"ssh": {
"command": "uv",
"args": ["run", "--directory", "/path/to/open-ssh-mcp", "ssh-mcp-server"],
"env": {
"SSH_MCP_CONFIG": "/path/to/config.json"
}
}
}
}
将
/path/to/替换为实际路径。Windows 用户请使用.exe文件路径。
4. 在其他 MCP 客户端中配置
Claude Desktop、Cursor 等客户端配置方式类似,在对应的 MCP 配置文件中添加上述 JSON 即可。
配置说明
配置为 JSON 格式,加载优先级:--config 参数 → SSH_MCP_CONFIG 环境变量 → 当前目录 config.json。
{
"defaults": {
"port": 22,
"username": "deploy",
"auth_method": "key",
"private_key_path": "~/.ssh/id_ed25519",
"known_hosts": "~/.ssh/known_hosts",
"connect_timeout_sec": 10,
"command_timeout_sec": 30,
"keepalive_interval_sec": 30,
"keepalive_count_max": 3
},
"servers": {
"web-prod-01": {
"host": "10.0.1.11",
"description": "生产 Web 服务器"
}
},
"security": {
"command_blocklist": ["rm -rf /", "mkfs", "shutdown", "reboot"],
"blocked_paths": ["/etc/shadow", "/etc/sudoers"],
"max_command_timeout_sec": 300,
"audit_log_path": "~/.ssh-mcp/audit.jsonl",
"require_confirmation_for": ["sudo", "systemctl restart", "rm"]
},
"pool": {
"max_connections": 10,
"idle_timeout_sec": 600,
"health_check_interval_sec": 60,
"auto_reconnect": true,
"max_reconnect_attempts": 3
}
}
设计要点:
defaults消除重复配置,每台服务器可覆盖任意字段password_env引用环境变量名 — 配置文件中不存储明文密码- 安全策略与服务器定义在同一文件中 — 单一配置源
可用工具
连接管理
| 工具 | 说明 |
|---|---|
ssh_connect |
建立 SSH 连接 |
ssh_disconnect |
断开连接 |
ssh_list_servers |
列出所有已配置服务器及连接状态 |
ssh_status |
检查连接健康状态:延迟、主机名、操作系统 |
命令执行
| 工具 | 说明 |
|---|---|
ssh_exec |
执行远程命令 |
ssh_exec_sudo |
以 sudo 执行(策略门控) |
ssh_exec_batch |
顺序执行多条命令 |
文件操作
| 工具 | 说明 |
|---|---|
ssh_upload |
上传本地文件到远程 |
ssh_download |
下载远程文件 |
ssh_read_file |
读取远程文件内容 |
ssh_write_file |
写入 / 追加远程文件 |
ssh_delete_file |
删除远程文件 / 目录 |
ssh_list_dir |
列出目录内容 |
ssh_stat |
获取文件元数据 |
部署运维
| 工具 | 说明 |
|---|---|
ssh_deploy |
完整部署:上传 → 备份 → 放置 → 重启 |
ssh_restart_service |
重启 systemd 服务 |
ssh_service_status |
查看 systemd 服务状态 |
ssh_service_logs |
获取近期服务日志 |
安全机制
- 命令黑名单 — 硬拒绝危险命令
- 路径限制 — 阻止访问敏感路径
- Sudo 门控 — 特权操作需确认
- 审计日志 — 每次工具调用记录到 JSONL(时间戳、工具、服务器、命令、结果、耗时)
- 主机密钥验证 — 默认通过
known_hosts启用
开发
# 安装开发依赖
pip install -e '.[dev]'
# 运行测试
pytest
# 代码检查
ruff check src/
# 本地打包(需要 PyInstaller)
pyinstaller --onefile --name open-ssh-mcp --paths src \
--collect-all pydantic_core --collect-all cryptography \
--collect-all asyncssh --collect-all mcp \
build_entry.py
许可证
MIT
Recommended Servers
playwright-mcp
A Model Context Protocol server that enables LLMs to interact with web pages through structured accessibility snapshots without requiring vision models or screenshots.
Magic Component Platform (MCP)
An AI-powered tool that generates modern UI components from natural language descriptions, integrating with popular IDEs to streamline UI development workflow.
Audiense Insights MCP Server
Enables interaction with Audiense Insights accounts via the Model Context Protocol, facilitating the extraction and analysis of marketing insights and audience data including demographics, behavior, and influencer engagement.
VeyraX MCP
Single MCP tool to connect all your favorite tools: Gmail, Calendar and 40 more.
graphlit-mcp-server
The Model Context Protocol (MCP) Server enables integration between MCP clients and the Graphlit service. Ingest anything from Slack to Gmail to podcast feeds, in addition to web crawling, into a Graphlit project - and then retrieve relevant contents from the MCP client.
Kagi MCP Server
An MCP server that integrates Kagi search capabilities with Claude AI, enabling Claude to perform real-time web searches when answering questions that require up-to-date information.
E2B
Using MCP to run code via e2b.
Neon Database
MCP server for interacting with Neon Management API and databases
Exa Search
A Model Context Protocol (MCP) server lets AI assistants like Claude use the Exa AI Search API for web searches. This setup allows AI models to get real-time web information in a safe and controlled way.
Qdrant Server
This repository is an example of how to create a MCP server for Qdrant, a vector search engine.