Skip to content

Latest commit

 

History

History

http-conversion

README.md

HTTP Endpoint Conversion Example

📹 Watch the demo video to see this example in action!

This example demonstrates gen-mcp's ability to automatically convert HTTP REST API endpoints into MCP tools. gen-mcp can expose any REST API as MCP tools that can be called by AI assistants, eliminating the need to write custom MCP server code.

Getting Started

1. Start the Feature Request API Server

First, run the Go server that provides the REST API:

cd feature-requests
go run main.go

The API will be available at http://localhost:9090 with endpoints:

  • GET /features - List all features (summaries only, sorted by upvotes)
  • GET /features/top - Get the highest-voted feature (summary only)
  • GET /features/{id} - Get detailed information about a specific feature
  • POST /features - Add a new feature request
  • POST /features/vote - Vote for a feature (increases upvotes)
  • POST /features/complete - Mark a feature request as completed
  • DELETE /features/{id} - Delete a feature request
  • GET /openapi.json - Get OpenAPI specification
  • POST /prompts/feature-analysis - Generate feature analysis prompt
  • GET /features/progress-report - Get feature progress report (static resource)

2. Generate Initial MCP Configuration

Use gen-mcp to automatically generate a starter configuration from the API:

genmcp convert http://localhost:9090/openapi.json

This creates two files:

  • mcpfile.yaml - Tool definitions (tools, prompts, resources)
  • mcpserver.yaml - Server configuration (runtime settings)

3. Customize the Configuration

Edit the generated files to:

  • MCP File (mcpfile.yaml): Select which endpoints should be exposed as MCP tools, improve tool descriptions, add usage instructions, configure input validation schemas
  • Server Config File (mcpserver.yaml): Configure runtime settings like port, logging, authentication

Example customizations in this demo:

  • Clear, specific descriptions for each tool
  • Guidance on when to call related tools (e.g., "Always call get_features-id after this tool...")
  • Proper input schemas with required parameters
  • Only exposing read endpoints initially (GET operations) for safety

4. Start the MCP Server

Launch the gen-mcp server with both files:

   genmcp run -f mcpfile.yaml -s mcpserver.yaml

The MCP server will run on port 8080 (as configured) and expose the HTTP endpoints as MCP tools that AI assistants can call seamlessly.

Key gen-mcp HTTP Conversion Features

  • Automatic Tool Generation: HTTP endpoints become MCP tools automatically from OpenAPI specs
  • Path Parameter Substitution: URL templates like {id} are handled automatically
  • Schema Validation: Input parameters are validated before API calls
  • Streamable HTTP Protocol: Real-time communication via streamablehttp
  • Flexible Configuration: Full control over which endpoints to expose and how
  • POST/PUT/DELETE Support: Can expose write operations like adding features, voting, completing, and deleting