MCP OpenAPI

MCP OpenAPI

This project aims to accelerate the adoption of the MCP protocol by simplifying server generation through automated tooling. It parses an OpenAPI specification and dynamically generates the corresponding MCP server components.

Key Features

• Parses OpenAPI specifications
• Dynamically generates MCP-compatible server code and tools
• Designed to streamline integration and deployment

Use Cases

One common use case is deploying this application as a sidecar alongside existing services. This enables seamless MCP support without requiring manual server implementation.

By automating the translation from OpenAPI to MCP, this tool reduces the overhead of manual server development and promotes faster adoption of the MCP protocol within modern service architectures.

Workflow

sequenceDiagram
    actor user as User
    actor agent as Agent
    participant ingress as Ingress
    participant mcp_proxy as MCP Proxy
    participant openapi_server as OpenAPI Server
    participant db as Database

    user->>ingress: Request for OpenAPI resource
    ingress->>openapi_server: Forward request
    openapi_server->>db: Query database
    db-->>openapi_server: Database response
    openapi_server-->> user: OpenAPI Server response

    Note over user,db: LLM interaction

    user->>agent: Send a prompt
    agent->>ingress: Request for MCP tool
    ingress->>mcp_proxy: Forward request
    mcp_proxy->>openapi_server: Forward request
    openapi_server->>db: Query database
    db-->>openapi_server: Database response
    openapi_server-->>mcp_proxy: OpenAPI Server response
    mcp_proxy-->>agent: MCP server response
    agent-->>user: Prompt response

Environment variables

Name	Default	Description
OPENAPI_SPEC_URL		URL of the Open API specification.
TRANSPORT	streamable-http	Transport type to use for communication with the MCP server.
HOST	127.0.0.1	Hostname or IP address to bind the server to.
PORT	8000	Port number to listen on.

How to setup a development environment

This project includes Dev Container support, making it easy to get started with a fully configured development environment. This ensures consistency across development setups and allows contributors to begin working without manual environment configuration.

Option 1: GitHub Codespaces

You can launch a cloud-based development environment instantly:

Option 2: Local Dev Container (Visual Studio Code)

1. Install Visual Studio Code.
2. Install the Dev Containers extension.
3. Ensure you have a container runtime installed and running:
   • Docker Desktop (Windows/macOS)
   • Docker Engine or Podman (Linux)
4. Open this repository in Visual Studio Code.
5. When prompted, reopen the project in a Dev Container.

Validation

A helper script is provided to validate the MCP server. It uses the MCP Python SDK to connect, list tools, and execute them.

List available tools

./tests/client.py -m http://localhost:8000/mcp

Example output:

{
  "properties": {
    "url": {
      "title": "Url",
      "type": "string"
    }
  },
  "required": ["url"],
  "title": "fetch_websiteArguments",
  "type": "object"
}

Call a tool

./tests/client.py -m http://localhost:8000/mcp -a call-tool -t fetch -i url=https://electrocucaracha.com/acerca/

Example output (truncated):

<!DOCTYPE html>
<html lang="es">
  <head>
    <meta charset="UTF-8" />
    ...
  </head>
</html>