Establishes a bridge between Model Context Protocol clients and local, OpenAI-compatible Large Language Model services.
Custom is a Model Context Protocol (MCP) server designed to facilitate seamless integration of MCP-compatible applications with your self-hosted, OpenAI-compatible Large Language Model (LLM) services. It translates MCP requests for various tasks, including chat completion, model listing, and health checks, to your local LLM. With robust features like streaming support, pre-built prompts for common AI tasks, and environment-based configuration, Custom serves as a vital component for developers looking to leverage local LLMs within the MCP ecosystem.
Key Features
01Integration with any OpenAI-compatible LLM service
02Supports chat completion, model listing, and health checks
03Full implementation of Model Context Protocol
040 GitHub stars
05Configurable via environment variables
06Provides both streaming and non-streaming responses
Use Cases
01Routing and managing AI requests to a self-hosted Large Language Model backend.
02Integrating local LLM services with MCP client applications like Claude Desktop.
03Developing and testing AI applications that utilize the Model Context Protocol standard.
