chat

This MCP server provides a cross‑platform desktop interface for interacting with multiple large language models (LLMs) through the Model Context Protocol (MCP). The app is built with Electron and serves as a testing ground to quickly configure and switch between different MCP backends and LLM providers, including OpenAI SDK compatible models and other MCP‑friendly endpoints. You can connect to one or more servers, manage multiple clients, and use the integrated UI to issue prompts, manage conversations, and evaluate response behavior across backends. The included configuration templates (such as gtp-api.json, qwen-api.json, and deepinfra.json) demonstrate how to point the UI at various MCP backends, including settings for model names, API URLs, paths, and whether MCP is enabled. This makes it straightforward to compare model performance and test multi‑server workflows from a single interface.

Prerequisites:

• Node.js and npm installed on your system
• A Git client (optional but recommended)

Installation steps:

1. Clone or download the repository:
   • git clone https://github.com/AI-QL/ai-ql-chat-mcp.git
   • cd ai-ql-chat-mcp
2. Install dependencies:
   • npm install
3. Build the Electron desktop application (optional for development):
   • npm run build-app
4. Run the application:
   • npm start

Notes:

• If you plan to customize MCP server configurations, you can edit the config files under src/main (for example, src/main/config.json) to ensure the command and paths point to valid executables or scripts on your machine.
• For Windows users, ensure PATH variables include Node.js and npm binaries; you may need to run the command prompt with administrator privileges for certain setup steps.

Tips and common issues:

• If the application has trouble starting due to missing binaries, verify that Node.js and npm are correctly installed by running node -v and npm -v in your terminal.
• When running multiple MCP servers, ensure each server has a unique name in the mcpServers map to avoid conflicts.
• The App can leverage various backends by editing the provided JSON templates (gtp-api.json, qwen-api.json, deepinfra.json). Update fields such as url, path, model, and apiKey as needed.
• If you encounter Electron download or build timeout issues, consider setting ELECTRON_MIRROR to a reachable mirror or increasing network reliability, as documented in the project troubleshooting section.
• For testing locally, you can use the example configuration snippet shown in the README to quickly switch between server backends and verify MCP interactions.