OpenRouter MCP Client

Example Use Cases

• 1

  Access multiple AI model providers (Google Gemini, Meta Llama, DeepSeek, etc.) through a single MCP endpoint.

• 2

  Orchestrate and combine outputs from several models (multi-model completion) to improve response quality or robustness.

• 3

  Tight integration with the Cursor editor via MCP, including model selection, caching of model metadata, and local deployment options.

Description

This project is an MCP client that integrates Cursor with OpenRouter.ai, enabling access to a wide range of models from multiple providers. It caches model metadata to reduce API calls, supports both free and paid models, and includes utilities for multi-model completion. The client can be run locally or deployed (e.g., to Smithery) and requires Node.js v18+ and an OpenRouter API key.