MedX

MedX MCP Server is an API-driven platform that lets AI-powered medical agents access diagnostic, decision-support, and personalized healthcare capabilities via a RESTful interface. It supports streaming results, asynchronous tool execution, and robust session management, making it easier to integrate AI-driven clinical workflows into your applications.

How to use

You interact with the MedX MCP Server through a client that communicates with the server’s REST API. Start by discovering the server’s capabilities, then call tools, and optionally stream results as they become available. You can manage conversation sessions, cancel in-flight actions, and verify health and readiness before issuing requests.

How to install

Prerequisites: Python 3.8+ and pip. You will run the server locally for development and testing.

1. Create a project directory and navigate into it.

2. Install server dependencies.

3. Set required environment variables for authentication and API access.

4. Start the MedX MCP Server.

Endpoints and basic usage

Discover server capabilities and tools via the manifest endpoint. Execute a tool with a targeted input, then stream results if you want real-time updates. If needed, cancel an ongoing call or check the health of the server.

Configuration and security

Environment variables you will configure include the OpenAI API key and the server authentication token. You should also be aware of host and port settings for local development.

Key environment variables shown in the setup include OpenAI API key, MCP server token, server host, and server port. These controls determine access to AI capabilities and the security of requests.

Examples and quick tips

A typical workflow involves discovering capabilities, issuing a call with a user message (for example, querying a medical concept), and streaming the response as it is generated. Use the stream endpoint for real-time updates and the cancel endpoint if you need to stop a long-running operation.

Notes on endpoints you will use

GET /mcp/manifest — Discover server capabilities and tools. POST /mcp/execute — Execute a tool. GET /mcp/stream/{call_id} — Stream results. POST /mcp/cancel/{call_id} — Cancel a call. GET /healthz — Health check. GET /readyz — Readiness check.

Troubleshooting

If you cannot connect, verify the server is running on the expected host and port, ensure required environment variables are set, and confirm the token used for authentication is valid. For streaming issues, check that the connection remains open and that the server is not dropping events.

Notes

The server is designed for asynchronous tool execution, session-based conversations, and robust health checks to support clinical workflows. Ensure your client handles streaming data and token-based authentication as described.

Available tools

discover_capabilities

Retrieve the server's capabilities and available tools via the manifest endpoint, including descriptions of supported features.

execute_tool

Execute a specified AI tool with given input to produce results.

stream_results

Stream results for an ongoing call so you receive data as it is produced.

cancel_call

Cancel an in-flight tool execution identified by call_id.

health_check

Perform a health check to verify the server is running and healthy.

readiness_check

Perform a readiness check to ensure the server is ready to handle requests.