rest-to -adapter

This MCP server provides a Python-based adapter that converts REST API specifications into MCP tools for AI agents. It ingests formats such as OpenAPI 3.x, Swagger 2.x, and OpenAPI Actions, normalizes endpoints, and generates MCP-compatible tool definitions that can be discovered and executed via the MCP protocol (JSON-RPC 2.0 over stdio). You can run the server to expose a tool registry that agents can list and call through the tools/list and tools/call endpoints, enabling agents to perform authenticated requests to the underlying REST APIs with built-in error handling and retry logic. The adapter supports built-in authentication handlers, automatic parameter filtering (hiding auth params from tool schemas), and a hybrid approach that blends defaults with auto-detection and overrides for precise control.

Prerequisites:

• Python 3.11+ installed on your system
• pip (comes with Python)
• Git (optional, for cloning the repository)

Step-by-step installation:

# Install the package from PyPI (stable release)
pip install rest-to-mcp-adapter

# Or install the latest beta features
pip install --pre rest-to-mcp-adapter

# Install a specific beta version
pip install rest-to-mcp-adapter==0.2.0b1

From source (for development):

git clone https://github.com/pawneetdev/rest-to-mcp-adapter.git
cd rest-to-mcp-adapter
pip install -e .

Dependencies installed automatically include:

• pydantic>=2.0.0
• pyyaml>=6.0
• requests>=2.31.0
• optional: langchain-community>=0.0.20 (if you need enhanced OpenAPI validation via extras: pip install rest-to-mcp-adapter[langchain])

Tips and considerations:

• The tool naming enforces a 64-character limit with truncation logic; plan endpoint/tool names accordingly.
• Auth handling is built-in; sensitive auth parameters are automatically hidden from tool schemas unless explicitly overridden.
• The MCP server exposes discovery at tools/list and supports execution via tools/call; you can integrate with Claude, GPT, or other agents that support MCP.
• When loading specifications, the library auto-detects input type (URL, file path, or raw content) and supports multiple formats (OpenAPI 3.x, Swagger 2.x, OpenAPI Actions).
• If you encounter timeouts or rate limits against target APIs, adjust REST_TO_MCP_ADAPTER_API_TIMEOUT and implement custom retry behavior as needed.
• For development, consider enabling verbose logging to diagnose issues with spec parsing or tool generation.