ChatMCP

What is chat-mcp?

chat-mcp is an open-source, cross-platform desktop chat application built with Electron that leverages the Model Context Protocol (MCP) to interact with various Large Language Models (LLMs). It aims to provide a clean, minimalistic codebase for understanding MCP principles and offers a quick way to test multiple LLM servers.

How to use chat-mcp?

To use chat-mcp, first clone or download the repository. Then, modify the config.json file in src/main to ensure valid command and path arguments. Ensure Node.js is installed. Install dependencies using npm install and start the application with npm start.

You can configure different LLM APIs (e.g., OpenAI, Qwen, DeepInfra) by creating JSON files with specific chatbotStore and defaultChoiceStore configurations. The application can also be built into a standalone desktop application using npm run build-app.

Key Features of chat-mcp

• Cross-Platform Compatibility: Supports Linux, macOS, and Windows.
• Flexible Apache-2.0 License: Allows modification and building of custom desktop applications.
• Dynamic LLM Configuration: Compatible with OpenAI SDK-supported LLMs, enabling quick testing of multiple backends.
• Multi-Client Management: Configure and manage multiple clients to connect to multiple servers using MCP config.
• UI Adaptability: The UI can be extracted for web use, ensuring consistent ecosystem and interaction logic.
• Multimodal Support: Demonstrated with images showing multimodal capabilities.
• Reasoning and Latex Support: Shown through examples in the README.
• MCP Tools Visualization & Toolcall Process Overview: Visual aids for understanding MCP functionality.
• MCP Prompts Template & Dynamic LLM Config: Features for managing prompts and LLM configurations.
• DevTool Troubleshooting: Built-in tools for debugging.

Use Cases of chat-mcp

• LLM Interaction: Seamlessly interact with various LLMs through a unified desktop interface.
• MCP Understanding: Provides a simplified codebase to understand the core principles of the Model Context Protocol.
• LLM Testing: Efficiently test multiple LLM servers and configurations.
• Developer Tool: Ideal for developers and researchers working with LLMs and MCP.
• Building Custom Applications: The flexible license allows for building custom desktop applications based on chat-mcp.

FAQ from chat-mcp

Q: How to resolve Error: spawn npx ENOENT? A: Modify the config.json in src/main. On Windows, npx might not work; consider using node in config.json with an absolute path to the server module. Ensure the provided path is valid.

Q: Why is npm install timing out? A: This often occurs due to slow or inaccessible Electron download servers. Modify the ELECTRON_MIRROR environment or global variable to use an accessible Electron mirror site.

Q: Why is electron-builder timing out? A: Unstable network connections can interrupt the download of large release packages from GitHub. On Windows, clear the cache in C:\Users\YOURUSERNAME\AppData\Local\electron and electron-builder. It is recommended to use the default shell terminal instead of VSCode's built-in terminal due to potential permission issues.

MCP Chat Desktop App

A Cross-Platform Interface for LLMs

This desktop application utilizes the MCP (Model Context Protocol) to seamlessly connect and interact with various Large Language Models (LLMs). Built on Electron, the app ensures full cross-platform compatibility, enabling smooth operation across different operating systems.

The primary objective of this project is to deliver a clean, minimalistic codebase that simplifies understanding the core principles of MCP. Additionally, it provides a quick and efficient way to test multiple servers and LLMs, making it an ideal tool for developers and researchers alike.

News

This project originated as a modified version of Chat-UI, initially adopting a minimalist code approach to implement core MCP functionality for educational purposes.

Through iterative updates to MCP, I received community feedback advocating for a completely new architecture - one that eliminates third-party CDN dependencies and establishes clearer modular structure to better support derivative development and debugging workflows.

This led to the creation of Tool Unitary User Interface, a restructured desktop application optimized for AI-powered development. Building upon the original foundation, TUUI serves as a practical AI-assisted development paradigm, if you're interested, you can also leverage AI to develop new features for TUUI. The platform employs a strict linting and formatting system to ensure AI-generated code adheres to coding standards.

📢 Update: June 2025
The current project refactoring has been largely completed, and a pre-release version is now available. Please refer to the following documentation for details:

• TUUI GitHub Repository
• TUUI Architecture
• TUUI Official Website

Features

• Cross-Platform Compatibility: Supports Linux, macOS, and Windows.

• Flexible Apache-2.0 License: Allows easy modification and building of your own desktop applications.

• Dynamic LLM Configuration: Compatible with all OpenAI SDK-supported LLMs, enabling quick testing of multiple backends through manual or preset configurations.

• Multi-Client Management: Configure and manage multiple clients to connect to multiple servers using MCP config.

• UI Adaptability: The UI can be directly extracted for web use, ensuring consistent ecosystem and interaction logic across web and desktop versions.

Architecture

Adopted a straightforward architecture consistent with the MCP documentation to facilitate a clear understanding of MCP principles by:

DeepWiki

How to use

After cloning or downloading this repository:

1. Please modify the config.json file located in src/main.
   Ensure that the command and path specified in the args are valid.

2. Please ensure that Node.js is installed on your system.
   You can verify this by running node -v and npm -v in your terminal to check their respective versions.

3. npm install

4. npm start

Configuration

Create a .json file and paste the following content into it. This file can then be provided as the interface configuration for the Chat UI.

• gtp-api.json

  {
      "chatbotStore": {
          "apiKey": "",
          "url": "https://api.aiql.com",
          "path": "/v1/chat/completions",
          "model": "gpt-4o-mini",
          "max_tokens_value": "",
          "mcp": true
      },
      "defaultChoiceStore": {
          "model": [
              "gpt-4o-mini",
              "gpt-4o",
              "gpt-4",
              "gpt-4-turbo"
          ]
      }
  }
  

You can replace the 'url' if you have direct access to the OpenAI API.

Alternatively, you can also use another API endpoint that supports function calls:

• qwen-api.json

  {
      "chatbotStore": {
          "apiKey": "",
          "url": "https://dashscope.aliyuncs.com/compatible-mode",
          "path": "/v1/chat/completions",
          "model": "qwen-turbo",
          "max_tokens_value": "",
          "mcp": true
      },
      "defaultChoiceStore": {
          "model": [
              "qwen-turbo",
              "qwen-plus",
              "qwen-max"
          ]
      }
  }
  

• deepinfra.json

  {
      "chatbotStore": {
          "apiKey": "",
          "url": "https://api.deepinfra.com",
          "path": "/v1/openai/chat/completions",
          "model": "meta-llama/Meta-Llama-3.1-70B-Instruct",
          "max_tokens_value": "32000",
          "mcp": true
      },
      "defaultChoiceStore": {
          "model": [
              "meta-llama/Meta-Llama-3.1-70B-Instruct",
              "meta-llama/Meta-Llama-3.1-405B-Instruct",
              "meta-llama/Meta-Llama-3.1-8B-Instruct"
          ]
      }
  }
  

Build Application

You can build your own desktop application by:

npm run build-app

This CLI helps you build and package your application for your current OS, with artifacts stored in the /artifacts directory.

For Debian/Ubuntu users experiencing RPM build issues, try one of the following solutions:

• Edit package.json to skip the RPM build step. Or

• Install rpm using sudo apt-get install rpm (You may need to run sudo apt update to ensure your package list is up-to-date)

Troubleshooting

Error: spawn npx ENOENT - ISSUE 40

Modify the config.json in src/main

On windows, npx may not work, please refer my workaround: ISSUE 101

• Or you can use node in config.json:

  {
      "mcpServers": {
          "filesystem": {
          "command": "node",
          "args": [
              "node_modules/@modelcontextprotocol/server-filesystem/dist/index.js",
              "D:/Github/mcp-test"
          ]
          }
      }
  }
  

Please ensure that the provided path is valid, especially if you are using a relative path. It is highly recommended to provide an absolute path for better clarity and accuracy.

By default, I will install server-everything, server-filesystem, and server-puppeteer for test purposes. However, you can install additional server libraries or use npx to utilize other server libraries as needed.

Installation timeout

Generally, after executing npm install for the entire project, the total size of files in the node_modules directory typically exceeds 500MB.

If the installation process stalls at less than 300MB and the progress bar remains static, it is likely due to a timeout during the installation of the latter part, specifically Electron.

This issue often arises because the download speed from Electron's default server is excessively slow or even inaccessible in certain regions. To resolve this, you can modify the environment or global variable ELECTRON_MIRROR to switch to an Electron mirror site that is accessible from your location.

Electron builder timeout

When using electron-builder to package files, it automatically downloads several large release packages from GitHub. If the network connection is unstable, this process may be interrupted or timeout.

On Windows, you may need to clear the cache located under the electron and electron-builder directories within C:\Users\YOURUSERNAME\AppData\Local before attempting to retry.

Due to potential terminal permission issues, it is recommended to use the default shell terminal instead of VSCode's built-in terminal.

Demo

Multimodal Support

Reasoning and Latex Support

MCP Tools Visualization

MCP Toolcall Process Overview

MCP Prompts Template

Dynamic LLM Config

DevTool Troubleshooting