Provides schema validation for function parameters, ensuring data correctness before processing API requests.
Offers containerized deployment of the MCP server, simplifying installation and providing consistent runtime environments.
Enables exposure of APIs compatible with the Model Context Protocol for use with OpenAI services, allowing custom functions to be invoked by AI agents.
mcp-starter
Secure, starter MCP server boilerplate with JWT authentication, schema validation, and Docker support.
Built for developers who want to expose AI-compatible APIs to tools like Claude, OpenAI, Sourcegraph Cody, or custom agents using the Model Context Protocol (MCP).
Features
JWT-based authentication with tenant isolation
Function auto-loading from
/functionsdirectoryInline per-function schema validation (via AJV)
Dynamic MCP manifest generation (
/mcp-manifest.json)Docker container
CLI utility to generate JWTs for local testing
Health endpoints (
/mcp,/healthz)
Related MCP server: MCP JSON-RPC Server
Running locally
Clone repo
Run with Docker
Run from code
Generating a Test Token
The --secret value (supersecure) must match the JWT_SECRET defined in your .env or docker-compose.yml.
Testing sample submitFeedback
You can play around with the body to remove message or give an invalid value for rating to see how validation is working.
Follow
Have feedback or feature ideas? Open an issue or contribute via pull request.