AI-QL/chat-mcp

Chat-MCP is a cross-platform desktop application designed to facilitate interaction with Large Language Models (LLMs) using the Model Context Protocol (MCP). Built on the Electron framework, it provides a clean and minimalistic codebase intended for educational purposes and for rapid testing of MCP servers and LLM backends. 0 license.

The architecture of Chat-MCP employs a three-process Electron model, separating responsibilities and maintaining security boundaries within the application. Its renderer process utilizes a centralized Pinia state management system, with specialized stores handling conversation arrays, user messages, LLM API calls, MCP server and tool management, endpoint configurations, conversation history, system prompts, workflow cards, prompt templates, resource templates, UI state, and notification messages. docx file reading for references.

A key feature of Chat-MCP is its support for any OpenAI-compatible API endpoint, configurable through JSON files. 1. Users can also define custom providers that conform to the OpenAI API specification. The application manages the full message flow, from adding user messages to conversations, initiating inference, executing tool calls via MCP servers, and aggregating tools from multiple MCP endpoints.

Chat-MCP is delivered as a desktop application compatible with Linux, macOS, and Windows. 0 license, making it suitable for developers, researchers, and educators who wish to explore MCP integration, test LLM workflows, or build upon its foundation for further development.