home / mcp / belaymit mcp server
Routes MCP requests to multiple downstream MCP servers and enables a LangGraph.js-based Dev Assistant with RAG integration.
Configuration
View docs{
"mcpServers": {
"belaymit-mcp": {
"url": "http://localhost:8001"
}
}
}You can route MCP requests from a central proxy to multiple downstream MCP servers, enabling a modular setup where data sources like filesystems, GitHub, Google Drive, and Atlassian services become accessible through a single MCP interface. This server setup lets you orchestrate requests, run a Dev Assistant with RAG capabilities, and test interactions end-to-end.
Start by running the MCP Proxy Server to expose downstream endpoints through a unified MCP URL. Then launch the Dev Assistant Agent to empower your MCP workflow with LangGraph.js and RAG capabilities. Finally, use the MCP Client Tester to validate interactions end-to-end and iterate with your IDE integration for a smooth developer experience.
Prerequisites: you need Node.js 18 or later, npm or yarn, and Git installed on your machine.
1. Clone the project repository and navigate into it.
git clone https://github.com/your-username/mcp.git
cd mcp2. Install dependencies for all MCP components.
npm install --legacy-peer-deps3. Build the project to compile TypeScript sources.
npm run build4. Run the MCP components you need. The proxy starts the main entry point and exposes the MCP URL to downstream servers.
npm run proxy:start
```
Starts on http://localhost:8002
npm run agent:start
npm run client:testThe MCP Proxy Server routes requests to several downstream MCP endpoints. There are four downstream targets available in this configuration: filesystem, github, gdrive, and atlassian. Each downstream target is accessed via HTTP and is configured in the proxy’s routing map.
[
{
"type": "http",
"name": "filesystem",
"url": "http://localhost:8001",
"args": []
},
{
"type": "http",
"name": "github",
"url": "http://localhost:8003",
"args": []
},
{
"type": "http",
"name": "gdrive",
"url": "http://localhost:8004",
"args": []
},
{
"type": "http",
"name": "atlassian",
"url": "http://localhost:8005",
"args": []
}
]
```
This configuration enables the proxy to forward requests to each downstream service at the specified localhost ports. Ensure each downstream service is running and reachable at its corresponding port before making requests through the proxy.Set environment variables to enable integrations and control behavior. Create a .env file in the project root with keys for API access, tokens, and proxy details as needed by your deployment. Common keys include OPENAI_API_KEY for LangChain integration, GITHUB_TOKEN for GitHub access, GOOGLE_DRIVE_CLIENT_ID and GOOGLE_DRIVE_CLIENT_SECRET for Google Drive access, and ATLASSIAN_API_TOKEN along with ATLASSIAN_INSTANCE_URL for Atlassian access. Include proxy host and port, and a log level to control verbosity.
If you encounter dependency conflicts, reinstall with the legacy peer-deps flag. If TypeScript errors occur, verify you are using TypeScript 5.5 or newer. Build failures often point to an incompatible Node.js version; ensure you are on Node.js 18+. If tests fail, confirm all required environment variables are set and downstream services are reachable.
You can development-test different components independently using the provided script aliases. Use the proxy:start command to verify routing, agent:start to validate the Dev Assistant behavior, and client:test to exercise the MCP client interactions. In development mode you can run npm run dev to watch for changes and automatically rebuild as you iterate.
Routes and forwards MCP requests to configured downstream MCP servers.
LangGraph.js-based agent that interacts with the MCP environment and supports RAG capabilities.
Integrates a retrieval-augmented generation flow to fetch context from the mock knowledge base and generate informed responses.
Tools to simulate MCP client interactions and validate server responses.
Configurations and guidance for integrating MCP support into IDEs like VS Code and Cursor.