mcpField
M
MCP Server SSH Server

MCP Server SSH Server

Local
github.com/Machine-To-MachineDeveloper Tools11Updated 69w ago

About MCP Server SSH Server

MCP Server SSH Server provides SSH access to remotely installed MCP tools, aggregating multiple MCP servers behind one interface. It supports key-based authentication and an optional HTTP API for key management.

Key features

  1. 01Remote MCP access over SSH
  2. 02Key-based client authentication
  3. 03Aggregate multiple MCP servers
  4. 04Configure servers with JSON
  5. 05Optional HTTP key management API
  6. 06Support prompts, resources, tools, and logging

Use cases

  1. 01Access MCP tools on a remote host
  2. 02Combine several MCP providers
  3. 03Register client keys through HTTP
  4. 04Configure a local MCP tool collection

MCP support

It proxies and merges configured MCP servers over SSH, exposing their prompts, resources, tools, and logging capabilities through a unified interface.

Configuration

Local server config is the same for every client. Copy it into your client’s MCP config file.

No install method listed yet

We haven’t collected a package or remote endpoint for this server. Follow the instructions in the project README.

Repository

Tools

No tool list yet

Local servers expose tools only once running on your machine, and the maintainer has not provided a tool list.

Details

From the project README

MCP Server SSH Server

PyPI version Python Versions License: MIT

A secure SSH server for accessing and interacting with MCP (Model Context Protocol) tools installed remotely. This server allows clients to connect to multiple MCP tool providers through a unified secure interface.

Features

  • Secure Remote Access: Access MCP tools remotely over SSH
  • Key-Based Authentication: Support for key-based authentication
  • Key Management API: Optional HTTP API for managing SSH keys
  • Server Aggregation: Proxy and merge multiple MCP servers into a unified interface
  • MCP Support: Compatible with MCP capabilities including:
    • Prompts
    • Resources
    • Tools
    • Logging
  • Dynamic Configuration: Configure available MCP servers through a simple JSON configuration

Authentication Flow

%%{init: {'theme':'default', 'themeVariables': { 'primaryColor': '#5D8AA8', 'primaryTextColor': '#fff', 'primaryBorderColor': '#1F456E', 'lineColor': '#5D8AA8', 'secondaryColor': '#006400', 'tertiaryColor': '#fff' }}}%%

sequenceDiagram
    participant Host as MCP Host<br>(Claude/Cursor)
    participant Client as MCP SSH Client
    participant KeySrv as Key Server<br>(HTTP API)
    participant SSHSrv as MCP SSH Server
    participant MCP as MCP Servers

    Note over Client,SSHSrv: Initial Key Exchange & Authentication
    
    Client->>Client: Generate SSH key pair<br>if does not exist
    Client->>KeySrv: GET /server_pub_key
    KeySrv->>Client: Return server's public key
    Client->>Client: Store server key in<br>temporary known_hosts
    Client->>KeySrv: POST /register<br>{client_pub_key: "ssh-ed25519 AAAA..."}
    KeySrv->>SSHSrv: Store client public key<br>in authorized keys
    KeySrv->>Client: {status: "success"}
    
    Note over Client,SSHSrv: Secure SSH Connection
    
    Client->>SSHSrv: SSH handshake with<br>client key authentication
    SSHSrv->>SSHSrv: Verify client key<br>against authorized keys
    SSHSrv->>Client: Authentication successful
    
    Note over Client,SSHSrv: MCP Communication
    
    Host->>Client: JSONRPC request
    Client->>SSHSrv: Forward request<br>over SSH tunnel
    SSHSrv->>MCP: Route request to<br>appropriate MCP server
    MCP->>SSHSrv: Process and return results
    SSHSrv->>Client: Send response over SSH
    Client->>Host: Return JSONRPC response
    
    Note over Client,SSHSrv: Session Management
    
    alt Session Termination
        Host->>Client: Close connection
        Client->>SSHSrv: Terminate SSH session
        SSHSrv->>SSHSrv: Clean up resources
    end

Installation

In a uv managed python project, add to dependencies by:

uv add m2m-mcp-server-ssh-server
# For key server functionality, include the optional dependency:
uv add m2m-mcp-server-ssh-server[key-server]

Alternatively, for projects using pip for dependencies:

pip install m2m-mcp-server-ssh-server
# Or with key server functionality
pip install m2m-mcp-server-ssh-server[key-server]

To install directly from the source:

git clone https://github.com/Machine-To-Machine/m2m-mcp-server-ssh-server.git
cd m2m-mcp-server-ssh-server
pip install -e .

To run the server inside your project:

uv run m2m-mcp-server-ssh-server

Common Use Cases

1. Local Development Environment

Quickly set up a local server with your preferred MCP tools:

  1. Create a configuration file (servers_config.json):
{
  "mcpServers": {
    "HackerNews": {
      "command": "uvx",
      "args": ["mcp-hn"]
    },
    "major-league-baseball": {
      "command": "uvx",
      "args": ["mcp_mlb_statsapi"]
    }
  }
}
  1. Start the server with the key management API for easier client connection:
uv run m2m-mcp-server-ssh-server --run-key-server
  1. Connect using the client:
uv run m2m-mcp-server-ssh-client --host localhost --port 8022 --use-key-server
2. Remote Hosting (Quick Setup)

To make your MCP tools available remotely:

  1. Install on your server:
uv add m2m-mcp-server-ssh-server[key-server]
  1. Create your configuration file with desired MCP tools

  2. Start the server binding to all interfaces:

uv run m2m-mcp-server-ssh-server --host 0.0.0.0 --run-key-server --key-server-host 0.0.0.0
  1. Connect from client machines:
uv run m2m-mcp-server-ssh-client --host your-server.example.com --port 8022 --use-key-server

For more detailed deployment instructions, see Cloud Deployment Guide.

Usage

Command Line Options
  • --host: SSH server host address to bind to (default: 127.0.0.1)
  • --port: SSH server port to listen on (default: 8022)
  • --authorized-clients: Path to authorized keys file (default: ~/.ssh/authorized_keys)
  • --server-key: Path to server private key file (default: ~/.ssh/m2m_mcp_server_ssh_server)
  • --passphrase: Passphrase for the private key (optional)
  • --servers-config: Path to server configurations JSON (default: servers_config.json)
  • --log-level: Set logging level (default: INFO)
  • --run-key-server: Run the HTTP key management server
  • --key-server-port: Port for the HTTP key management server (default: 8000)
  • --key-server-host: Host for the HTTP key management server (default: 127.0.0.1)
  • --version: Show version information and exit

Security Note: When binding to all interfaces (0.0.0.0), your server will be accessible from any machine that can reach your system. Make sure to use strong authentication and consider additional security measures like firewalls.

Key Management Server

You can optionally run a key management HTTP server alongside the SSH server to simplify client key registration:

uv run m2m-mcp-server-ssh-server --run-key-server

This starts:

  • SSH server on port 8022
  • Key management HTTP server on port 8000 (localhost only by default)
Key Management API

The key management server provides the following endpoints:

  1. POST /register - Register a client's public key

    • Request body: {"client_pub_key": "ssh-rsa AAAA..."}
    • Response: {"status": "success"} or error message
  2. GET /server_pub_key - Get the server's public key

    • Response: {"server_pub_key": "ssh-ed25519 AAAA..."}
  3. GET /health - Check server health status

    • Response: {"status": "healthy"}

Example client key registration:

curl -X POST -H "Content-Type: application/json" \
  -d '{"client_pub_key": "ssh-rsa AAAA..."}' \
  http://localhost:8000/register

Security Considerations

  • Default binding is to localhost (127.0.0.1) for improved security
  • Keys are automatically generated with secure permissions if not provided
  • Default key location is ~/.ssh/m2m_mcp_server_ssh_server
  • Private keys use 0o600 permissions (readable only by owner)
  • Public keys use 0o644 permissions (readable by all, writable by owner)
  • Key server uses rate limiting for API endpoints
  • Input validation protects against malicious inputs

Dependencies

  • anyio (>=4.5.0)
  • asyncssh (>=2.20.0)
  • mcp (>=1.1.3)
  • pydantic (>=2.0.0)
  • aiohttp (>=3.11.16) - Optional, for key server functionality

Development

Setup Development Environment
git clone https://github.com/Machine-To-Machine/m2m-mcp-server-ssh-server.git
cd m2m-mcp-server-ssh-server
uv venv
source .venv/bin/activate  # On Windows: .venv\Scripts\activate
uv pip install -e ".[dev]"
Code Quality
# Run linting
uv run ruff check .

# Run formatting check
uv run ruff format --check .

# Run security checks
uv run bandit -r src/

Contributing

We welcome contributions to help expand and improve m2m-mcp-server-ssh-server. Whether you want to add new features, enhance existing functionality, or improve documentation, your input is valuable.

Pull requests are welcome! Feel free to contribute new ideas, bug fixes, or enhancements.

Contribution Guidelines
  1. Fork the repository
  2. Create a feature branch: git checkout -b feature-name
  3. Commit your changes: git commit -am 'Add some feature'
  4. Push to the branch: git push origin feature-name
  5. Submit a pull request

License

MIT License - See LICENSE file for details.

Authors

  • Machine To Machine

Comments

Markdown supported

New to MCP?

From what the protocol is to installing your first server, these answers cover most first questions.

An open protocol that defines how AI applications connect to external tools and data in a uniform way. Clients talk to the model; MCP servers provide tools, resources and prompt templates. They exchange standard messages, so one server can be reused by many clients.