Skip to main content
Glama
InfraMCP
by InfraMCP
README.md
# SSH MCP Server

[![Python 3.10+](https://img.shields.io/badge/python-3.10%2B-blue.svg)](https://www.python.org/downloads/)
[![License: MIT](https://img.shields.io/badge/License-MIT-yellow.svg)](https://opensource.org/licenses/MIT)
[![Pylint](https://github.com/rorymcmahon/ssh-mcp-server/actions/workflows/pylint.yml/badge.svg)](https://github.com/rorymcmahon/ssh-mcp-server/actions/workflows/pylint.yml)
[![Safety Security Scan](https://github.com/rorymcmahon/ssh-mcp-server/actions/workflows/safety-scan.yml/badge.svg)](https://github.com/rorymcmahon/ssh-mcp-server/actions/workflows/safety-scan.yml)
[![Dependency Security Check](https://github.com/rorymcmahon/ssh-mcp-server/actions/workflows/dependency-security.yml/badge.svg)](https://github.com/rorymcmahon/ssh-mcp-server/actions/workflows/dependency-security.yml)

An MCP (Model Context Protocol) server that provides SSH client functionality for remote Linux server management. This server enables AI assistants to execute commands on remote Linux hosts via SSH, solving the limitations of built-in tools when working with remote systems.

## Why SSH MCP Server?

Built-in MCP tools are limited to local operations. This server extends AI capabilities to remote Linux systems by providing:

- **Remote Command Execution**: Execute any command on remote Linux hosts
- **System Administration**: Manage services, check system health, monitor processes
- **Secure Authentication**: Multiple secure credential storage options
- **Enterprise Integration**: Works with domain-joined systems and enterprise environments

## Features

- **SSH Command Execution**: Execute arbitrary commands on remote Linux hosts
- **Sudo Support**: Run commands with elevated privileges (secure password handling)

## Installation

### From PyPI (Recommended)

```bash
pip install ssh-mcp-server
```

### From Source

```bash
git clone https://github.com/rorymcmahon/ssh-mcp-server.git
cd ssh-mcp-server
pip install -e .
```
- **System Information**: Get system stats, processes, disk usage, and services
- **Secure Credentials**: Secure credential storage (currently macOS Keychain, expanding to other providers)
- **Connection Management**: Automatic connection handling with timeouts
- **Error Handling**: Comprehensive error reporting and recovery
- **Puppet Integration**: Run Puppet agent in no-op mode for configuration management

## Installation

### From PyPI (when published)
```bash
pip install ssh-mcp-server
```

### From Source
```bash
git clone https://github.com/rorymcmahon/ssh-mcp-server.git
cd ssh-mcp-server
pip install -e .
```

### Development Installation
```bash
git clone https://github.com/rorymcmahon/ssh-mcp-server.git
cd ssh-mcp-server
pip install -e ".[dev]"
```

## Configuration

### MCP Client Configuration

Add to your MCP client configuration (e.g., Claude Desktop, Q CLI):

```json
{
  "mcpServers": {
    "ssh": {
      "command": "ssh-mcp-server",
      "args": []
    }
  }
}
```

## Available Tools

### Core SSH Operations

#### `execute_ssh(hostname: str, command: str)`
Execute a command on a remote Linux host via SSH.

**Parameters:**
- `hostname`: Target hostname (e.g., "server.company.local")
- `command`: Command to execute

**Returns:**
```json
{
  "status": 0,
  "stdout": "total 24\ndrwxr-xr-x 3 user user 4096 ...",
  "stderr": ""
}
```
Or on error:
```json
{
  "error": "SSH connection or authentication failed"
}
```

#### `execute_sudo(hostname: str, command: str)`
Execute a command with sudo privileges. Automatically handles password input securely.

**Returns:** Same format as `execute_ssh`

### System Information Tools

#### `ssh_get_system_info(hostname: str)`
Get basic system information (OS, kernel, memory, root disk usage).

#### `get_running_processes(hostname: str)`
Get top 10 CPU-consuming processes.

#### `get_disk_usage(hostname: str)`
Get disk usage for all mounted filesystems.

#### `get_services(hostname: str)`
Get top 20 running systemd services.

#### `ssh_puppet_noop(hostname: str)`
Run Puppet agent in no-op mode (dry run) with verbose output.

## Usage Examples

### Basic Command Execution
```python
# Execute a simple command
result = execute_ssh("server.company.local", "uptime")
if "error" not in result:
    print(result["stdout"])  # System uptime information
```

### System Administration
```python
# Check system health
system_info = ssh_get_system_info("server.company.local")
disk_usage = get_disk_usage("server.company.local")
processes = get_running_processes("server.company.local")

# Restart a service with sudo
result = execute_sudo("server.company.local", "systemctl restart nginx")
```

### Error Handling
```python
result = execute_ssh("server.company.local", "invalid_command")
if "error" in result:
    print(f"Error: {result['error']}")
elif result["status"] != 0:
    print(f"Command failed with exit code {result['status']}")
    print(f"Error output: {result['stderr']}")
```

## Security Considerations

- **Credential Storage**: Uses secure credential storage (Keychain, future: Vault, AWS Secrets Manager)
- **Network Security**: Ensure SSH connections are over secure networks
- **Access Control**: Limit SSH user permissions on target hosts
- **Audit Logging**: Monitor SSH access and command execution
- **TouchID Protection**: macOS Keychain integration requires TouchID/password for access
- **Password Security**: Sudo passwords are passed securely via stdin, not visible in process lists

## Development

### Running Tests
```bash
pytest
```

### Code Formatting
```bash
black src/ tests/
isort src/ tests/
```

### Type Checking
```bash
mypy src/
```

### Coverage Report
```bash
pytest --cov=ssh_mcp_server --cov-report=html
```

## Contributing

1. Fork the repository
2. Create a feature branch (`git checkout -b feature/amazing-feature`)
3. Make your changes
4. Add tests for new functionality
5. Ensure all tests pass (`pytest`)
6. Format code (`black` and `isort`)
7. Commit changes (`git commit -m 'Add amazing feature'`)
8. Push to branch (`git push origin feature/amazing-feature`)
9. Open a Pull Request

## Roadmap

- [ ] SSH key-based authentication
- [ ] AWS Secrets Manager credential provider
- [ ] HashiCorp Vault credential provider
- [ ] Azure Key Vault credential provider
- [ ] Connection pooling and reuse
- [ ] File transfer operations (SCP/SFTP)
- [ ] Interactive shell sessions
- [ ] Connection health monitoring
- [ ] Batch command execution
- [ ] Custom SSH client configuration
- [ ] Windows support (additional credential providers)

## License

This project is licensed under the MIT License - see the [LICENSE](LICENSE) file for details.

## Support

- **Issues**: [GitHub Issues](https://github.com/rorymcmahon/ssh-mcp-server/issues)
- **Discussions**: [GitHub Discussions](https://github.com/rorymcmahon/ssh-mcp-server/discussions)
- **Changelog**: [CHANGELOG.md](CHANGELOG.md)

TDQS

C2.5/5.0

Scored across 15 tools

Disambiguation2/5

Five of fifteen tools are explicit 'Legacy compatibility' duplicates of newer tools with near-identical descriptions (get_running_processes vs ssh_get_running_processes, execute_ssh vs ssh_execute_ssh, execute_sudo vs ssh_execute_sudo, cache_credentials vs ssh_setup_credentials). An agent must guess whether the legacy or ssh_-prefixed variant is the intended entry point, and the descriptions offer no behavioral difference to disambiguate.

Naming Consistency2/5

All names are snake_case, but two competing prefix schemes coexist (bare verb_noun vs ssh_-prefixed), and verb ordering flips between pairs like execute_ssh and ssh_execute_ssh. The result is a set where the same action is spelled two different ways, which undermines predictability.

Tool Count3/5

Fifteen tools is within a reasonable range for an SSH/remote-admin server, and the functional surface (execute, sudo, credentials, system info, processes, disk, services, puppet) is sensibly scoped. However, roughly a third of the count is pure legacy duplication, so the effective tool set is ~10 padded out with redundant entries.

Completeness4/5

Core remote-admin workflows are covered: command execution, privileged execution, credential lifecycle (setup/clear/cache), system info, processes, disk, services, and Puppet dry-run. Gaps exist around file transfer/scp and host inventory listing, but agents can work around these via the execute tools.

Maintenance

ActivityMaintained
ResponsivenessNo issues