Unichat MCP Server

Example Use Cases

• 1

  Route requests to different LLM vendors (OpenAI, MistralAI, Anthropic, xAI, Google AI, DeepSeek) through a single MCP endpoint.

• 2

  Run in local or desktop environments using STDIO or SSE transports for integration with tools like Claude Desktop or custom clients.

• 3

  Built-in developer-oriented prompts (code_review, document_code, explain_code, code_rework) and a simple 'unichat' tool to standardize chat requests to downstream models.

Description

This project implements an MCP (Model Context Protocol) server in TypeScript that forwards requests to various LLM vendors using a single unified interface. It supports both STDIO and SSE transports and includes a built-in 'unichat' tool plus several predefined prompts (code_review, document_code, explain_code, code_rework). The server is easy to run locally or via npx and requires the vendor API key and a chosen model to operate, making it useful for integrating multiple model providers into agent workflows.