openapi-mcp-proxy

OpenAPI MCP Server

An MCP server that provides tools for exploring large OpenAPI schemas without loading entire schemas into LLM context. Perfect for discovering and analyzing endpoints, data models, and API structure efficiently.

Features

• API Configuration Management: Save and manage multiple API configurations with authentication headers if needed
• Schema Caching: Automatic caching of OpenAPI schemas to avoid repeated downloads
• Endpoint Discovery: List and search through API endpoints
• Pagination Support: Handle large APIs efficiently with configurable page sizes
• Detailed Schema Exploration: Get comprehensive information about endpoints and data models
• Efficient Context Usage: Explore large APIs without overwhelming LLM context windows

Prerequisites

• Python 3.13+: The server requires Python 3.13 or later
• uv: Fast Python package installer and resolver (installation guide)
• MCP-compatible client: Claude Desktop, Claude Code CLI, Cursor, or other MCP clients

Installing uv

macOS/Linux:

Using pip:

Installation

1. Clone the repository:

2. Install dependencies:

3. Verify installation:

The server should start without errors.

Usage

Running the Server

The server runs using stdio and integrates with MCP-compatible LLM clients.

Available Tools

API Management

• add_api: Add a new API configuration with name, URL and optional description
  • name (required): Short name for the API
  • url (required): URL to the OpenAPI scheme (yaml/json)
  • description (optional): Optional description
  • headers (optional): Optional HTTP headers for authentication (e.g., {'Authorization': 'Bearer token', 'X-API-Key': 'key'})
• list_saved_apis: List all saved API configurations
• remove_api: Remove a saved API configuration

API Exploration

• get_api_info: Get general information about an API
• list_endpoints: List all endpoints in an API with pagination and filtering
• search_endpoints: Search endpoints by query with pagination and filtering
• get_endpoint_details: Get detailed information about a specific endpoint
• list_models: List all data models in an API with pagination and filtering
• get_model_schema: Get detailed schema for a specific model

Tools Capabilities

Pagination

All listing tools (list_endpoints, search_endpoints, list_models) support pagination to handle large APIs efficiently:

• Default page size: 50 items
• Responses include navigation information (current page, total pages, has next/previous)

Advanced Filtering

Tools are capable to filter results to find exactly what you need:

Endpoint Filtering:

• HTTP methods (GET, POST, PUT, DELETE, etc.)
• Tags (include/exclude specific tags)
• Authentication requirements
• Deprecation status

Model Filtering:

• Model types (object, array, string, etc.)
• Property count (min/max number of properties)
• Required fields presence
• Tags (include/exclude specific tags)

Configuration

API configurations are automatically saved to api_configs.json in the working directory. The file structure: