MCP Remote Server
Enables execution of SSH commands on remote servers and management of Google Cloud Platform (GCE) instances through Cursor IDE.
README
MCP Remote Server
A Model Context Protocol (MCP) server that enables Cursor IDE to execute SSH commands on remote servers and manage Google Cloud Platform (GCE) instances. This server provides a seamless integration between Cursor and your remote infrastructure.
Features
- 🔐 SSH Command Execution: Run commands on remote servers via SSH
- ☁️ GCE Instance Management: Start and stop Google Cloud Compute Engine instances
- 🔄 Connection Pooling: Reuses SSH connections for improved performance
- 📝 SSH Config Integration: Uses your existing
~/.ssh/configfor host configuration - 🚀 MCP Protocol: Full support for Model Context Protocol v2025-06-18
Prerequisites
- Node.js 16+ installed
- SSH access configured to your remote server
- Google Cloud Platform credentials (for GCE features)
- Cursor IDE installed
Installation
-
Clone or navigate to the repository:
cd mcp-remote -
Install dependencies:
npm install -
Start the MCP server:
node src/index.jsThe server will start on
http://localhost:3000by default.
Configuration
1. SSH Configuration
First, you need to configure your SSH connection in ~/.ssh/config. The server always uses the host alias dev regardless of what you specify in prompts.
Create or edit ~/.ssh/config:
Host dev
HostName your-server.example.com
User your-username
Port 22
IdentityFile ~/.ssh/id_rsa
# Optional: Add additional SSH options
# ServerAliveInterval 60
# ServerAliveCountMax 3
# StrictHostKeyChecking no
# UserKnownHostsFile ~/.ssh/known_hosts
Example with AWS EC2:
Host dev
HostName ec2-12-34-56-78.compute-1.amazonaws.com
User ubuntu
Port 22
IdentityFile ~/.ssh/my-aws-key.pem
Example with Google Cloud:
Host dev
HostName 34.123.45.67
User root
Port 22
IdentityFile ~/.ssh/gcp-key.pem
Important Notes:
- The host alias must be named
dev(all lowercase) - The server will always connect to
devregardless of the host you specify in prompts - Make sure your SSH key has proper permissions:
chmod 600 ~/.ssh/id_rsa
2. Cursor MCP Configuration
Configure Cursor to connect to the MCP server by editing ~/.cursor/mcp.json:
{
"mcpServers": {
"local-mcp": {
"name": "Local MCP",
"url": "http://localhost:3000",
"description": "Runs SSH commands and manages GCE instances",
"apiKeyHeader": "x-api-key",
"apiKeyValue": "8BOe6KshR5Eq0otTSjSGMEf5P1edKcq"
}
}
}
Configuration Fields:
name: Display name for the MCP server in Cursorurl: The URL where your MCP server is running (default:http://localhost:3000)description: Description of what the server doesapiKeyHeader: Header name for API authentication (currently optional)apiKeyValue: API key value for authentication (currently optional)
After modifying mcp.json:
- Save the file
- Restart Cursor IDE
- The MCP server should appear in Cursor's tool list
3. Environment Variables (Optional)
You can set environment variables to customize behavior:
# SSH Configuration (optional - defaults to ~/.ssh/id_rsa)
export SSH_USERNAME=your-username
export SSH_PRIVATE_KEY_PATH=~/.ssh/custom_key
# Server Configuration
export PORT=3000
# GCP Configuration (for GCE features)
export GOOGLE_APPLICATION_CREDENTIALS=/path/to/gcp-credentials.json
Usage
Starting the Server
# Start the server
node src/index.js
# Or with custom port
PORT=8080 node src/index.js
The server will output:
MCP server running on http://localhost:3000
Ready to accept MCP connections from Cursor
Using in Cursor IDE
Once configured, you can use the MCP tools directly in Cursor:
-
Run SSH Commands:
- Ask Cursor: "Check disk usage on server1.example.com"
- The server will automatically connect to
devand execute the command - Example: "Run
df -hon my remote server"
-
Manage GCE Instances:
- Ask Cursor: "Start instance my-vm in project my-project, zone us-central1-a"
- Ask Cursor: "Stop instance my-vm in project my-project, zone us-central1-a"
Available Tools
-
run_command
- Execute SSH commands on the remote server
- Always connects to
devfrom~/.ssh/config - Parameters:
host: Ignored (always usesdev)cmdKey: The command to execute (e.g.,df -h,ls -la)
-
start_instance
- Start a GCE instance
- Parameters:
project: GCP project IDzone: GCE zone (e.g.,us-central1-a)instance: Instance name
-
stop_instance
- Stop a GCE instance
- Parameters:
project: GCP project IDzone: GCE zoneinstance: Instance name
Architecture
Connection Pooling
The server maintains a pool of SSH connections to improve performance:
- Connections are reused across multiple commands
- Connections are kept alive with keepalive settings
- Automatic cleanup on connection errors
SSH Config Integration
The server reads connection details from ~/.ssh/config:
- Hostname, port, username, and identity file
- Supports all standard SSH config options
- Handles
~expansion in file paths - Supports multiple IdentityFile entries (uses the first one)
Troubleshooting
Server Won't Start
-
Check if port is already in use:
lsof -i :3000 -
Check Node.js version:
node --version # Should be 16+
Cursor Can't Connect
-
Verify server is running:
curl http://localhost:3000 -
Check
~/.cursor/mcp.jsonsyntax:- Ensure valid JSON
- Verify the URL matches your server
-
Restart Cursor IDE after modifying
mcp.json
SSH Connection Issues
-
Test SSH connection manually:
ssh dev -
Verify SSH config:
ssh -F ~/.ssh/config dev -
Check SSH key permissions:
ls -la ~/.ssh/id_rsa chmod 600 ~/.ssh/id_rsa -
Verify the host alias exists:
grep -A 5 "^Host dev" ~/.ssh/config
Commands Not Executing
- Check server logs for error messages
- Verify the
devhost is properly configured in~/.ssh/config - Check SSH key has access to the remote server
Development
Project Structure
mcp-remote/
├── src/
│ ├── index.js # Main MCP server
│ ├── ssh.js # SSH connection handling
│ └── gcp.js # GCP instance management
├── package.json
├── README.md
└── .gitignore
Adding New Tools
To add new MCP tools:
- Add tool definition to the
toolsarray insrc/index.js - Add handler in the
tools/callcase - Implement the functionality in the appropriate module
Testing
# Test SSH connection
node -e "const {runRemoteCommand} = require('./src/ssh'); runRemoteCommand({host: 'dev', command: 'echo test'}).then(r => console.log(r));"
# Test MCP server
curl -X POST http://localhost:3000 \
-H "Content-Type: application/json" \
-d '{"jsonrpc":"2.0","method":"initialize","params":{"protocolVersion":"2025-06-18"},"id":1}'
Security Considerations
- The server runs on
localhost:3000by default (not exposed to network) - SSH credentials are read from
~/.ssh/config(standard SSH security) - API key authentication is available but currently optional
- Consider using environment variables for sensitive data
License
[Add your license here]
Contributing
Contributions are welcome! Please feel free to submit a Pull Request.
Support
For issues and questions:
- Check the Troubleshooting section
- Review server logs for error messages
- Ensure SSH configuration is correct
Note: The server always connects to the dev host alias from your SSH config, regardless of the host you specify in prompts. Make sure dev is properly configured in ~/.ssh/config.
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.