About MCP Server Runner
MCP Server Runner is a WebSocket server for running Model Context Protocol (MCP) servers and exposing them over WebSocket connections. It targets developers who need to reach MCP servers from web applications or other network-connected clients. The project is still work in progress, so its APIs may change.
Key features
- 01Access MCP servers over WebSocket connections
- 02Launch and manage MCP server processes
- 03Bidirectional message passing between client and server
- 04Graceful shutdown and error logging
- 05Multiple server definitions via JSON configuration
- 06Docker deployment support
Use cases
- 01Connecting a local MCP server to a web application
- 02Letting network clients reach an MCP server
- 03Deploying an MCP server in a container
MCP support
The project bridges WebSocket clients and MCP server implementations, launching the MCP server process and forwarding messages in both directions.
Configuration
Local server config is the same for every client. Copy it into your client’s MCP config file.
We haven’t collected a package or remote endpoint for this server. Follow the instructions in the project README.
Tools
Local servers expose tools only once running on your machine, and the maintainer has not provided a tool list.
Details
From the project READMEMCP Server Runner
Note: This project is currently under active development and in WIP (Work In Progress) status. Features and APIs may change significantly.
A WebSocket server implementation for running Model Context Protocol (MCP) servers. This application enables MCP servers to be accessed via WebSocket connections, facilitating integration with web applications and other network-enabled clients.
Development Status
- 🚧 Work In Progress: This software is in active development
- ⚠️ API Stability: APIs and features may change without notice
- 🧪 Testing: Currently undergoing testing and refinement
- 📝 Documentation: Documentation is being actively updated
Overview
MCP Server Runner acts as a bridge between WebSocket clients and MCP server implementations. It:
- Launches an MCP server process
- Manages WebSocket connections
- Handles bidirectional communication between clients and the MCP server
- Supports graceful shutdown and error handling
Features
- WebSocket server implementation with single-client support
- Process management for MCP server instances
- Bidirectional message passing between client and server
- Graceful shutdown handling
- Comprehensive error logging
- Cross-platform support (Unix/Windows)
Prerequisites
- Rust 1.70 or higher
- An MCP server implementation executable
Configuration
Environment Variables
The application can be configured through environment variables:
PROGRAM= # Path to the MCP server executable (required if no config file)
ARGS= # Comma-separated list of arguments for the MCP server
HOST=0.0.0.0 # Host address to bind to (default: 0.0.0.0)
PORT=8080 # Port to listen on (default: 8080)
CONFIG_FILE= # Path to JSON configuration file
Additional environment variables will be passed through to the MCP server process.
JSON Configuration
Alternatively, you can provide a JSON configuration file:
{
"servers": {
"filesystem": {
"command": "npx",
"args": [
"-y",
"@modelcontextprotocol/server-filesystem",
"/path/to/workspace"
]
},
"github": {
"command": "npx",
"args": ["-y", "@modelcontextprotocol/server-github"],
"env": {
"GITHUB_PERSONAL_ACCESS_TOKEN": "your_token_here"
}
}
},
"default_server": "filesystem",
"host": "0.0.0.0",
"port": 8080
}
You can specify the configuration file in two ways:
- As a command-line argument:
mcp-server-runner config.json - Using the
CONFIG_FILEenvironment variable:CONFIG_FILE=config.json mcp-server-runner
The JSON configuration allows you to define multiple server configurations and select one as the default.
Configuration Priority
- Command-line specified config file
CONFIG_FILEenvironment variable- Environment variables (
PROGRAM,ARGS, etc.) - Default values
Usage
-
Using environment variables:
export PROGRAM=npx export ARGS=-y,@modelcontextprotocol/server-github export PORT=8080 export GITHUB_PERSONAL_ACCESS_TOKEN=github_pat_*** cargo run -
Using a configuration file:
# Either specify the config file as an argument cargo run config.json # Or use the CONFIG_FILE environment variable CONFIG_FILE=config.json cargo run -
Connect to the WebSocket server:
const ws = new WebSocket("ws://localhost:8080");
Docker Support
A Dockerfile and docker-compose.yml are provided for containerized deployment:
docker-compose up --build
Development
Build the project:
cargo build
Run tests:
cargo test
Run with debug logging:
RUST_LOG=debug cargo run
Architecture
The application follows a modular architecture:
main.rs: Application entry point and server setupconfig/: Configuration loading and managementprocess/: Process management and I/O handlingwebsocket/: WebSocket connection managementstate.rs: Global state managementshutdown.rs: Graceful shutdown handling
Error Handling
- Standard error output from the MCP server is logged but not forwarded to clients
- WebSocket connection errors are handled gracefully
- Process errors are logged with detailed information
Limitations
- Supports only one client connection at a time
- Does not support WebSocket SSL/TLS (use a reverse proxy for secure connections)
- No built-in authentication mechanism
Contributing
- Fork the repository
- Create your feature branch (
git checkout -b feature/amazing-feature) - Commit your changes (
git commit -m 'Add amazing feature') - Push to the branch (
git push origin feature/amazing-feature) - Open a Pull Request
License
This project is licensed under the MIT License - see the LICENSE file for details.
Additional Resources
Comments
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.