BigSweetPotatoStudio/HyperChat

中文 | English

Features

HyperChat is an open-minded Chat client that can use various LLM APIs, fully supports MCP, and provides the best chat experience. It also implements the MAX productivity tools based on native agents.

• Supports OpenAI-style LLMs, OpenAI, Claude(OpenRouter), Qwen, Deepseek, GLM, Ollama.
• Built-in MCP plugin market with user-friendly MCP installation and configuration, one-click installation, welcome to submit HyperChatMCP.
• Also supports manual installation of third-party MCPs, simply fill in command, args, and env.

MCP:

• MCP extension market + third-party MCP support
• Supports freely creating Agent, can preset prompts and select MCP functions
• 🪟Windows + 🍏MacOS
• Supports dark mode 🌙
• Resources, Prompts, and Tools support
• Supports English and Chinese
• Supports Artifacts, SVG, HTML rendering, JavaScript error catching, supports opening Chrome's console
• Agent display optimization, supports search and drag-and-drop sorting
• Supports KaTeX, displays mathematical formulas, code rendering with highlighting and quick copy
• WebDAV synchronization
• Adds RAG, based on the MCP knowledge base
• Introduces the concept of ChatSpace, supports simultaneous chatting in multiple conversations
• Supports Agent Call Agent through HyperAgent's MCP
• Adds scheduled tasks to specify Agents to complete tasks on time, and view task completion status.

TODO:

• Will later add a docker version, built-in Linux desktop, built-in Chrome configuration to remote port, unified environment for easier handling. Then control via web interface. Can be used on any device, including mobile phones 🤣
• Permission pop-up, whether to allow or not
• Implement the ability for LLM to create its own MCP

LLM

LLM	Usability	Notes
claude	⭐⭐⭐⭐⭐	No explanation
openai	⭐⭐⭐⭐⭐	Perfectly supports multi-step function calls (gpt-4o-mini also works)
gemini flash 2.0	⭐⭐⭐⭐🌙	Very user-friendly
qwen	⭐⭐⭐⭐🌙	Very user-friendly
doubao	⭐⭐⭐🌙🌙	Feels okay to use
deepseek	⭐⭐⭐🌙🌙	Multi-step function calls may have issues

Usage

• 1. Configure the APIKEY, ensure your LLM service is compatible with OpenAI style.
• 2. Make sure uv + nodejs is installed on your system.

uvx & uv

Install using command line, or check the official Github tutorial uv

# MacOS
brew install uv
# Windows
winget install --id=astral-sh.uv  -e

npx & nodejs

Install using command line, or go to the official site to download and install, official site nodejs

# MacOS
brew install node
# Windows
winget install OpenJS.NodeJS.LTS

Development

cd electron && npm install
cd web && npm install
npm install
npm run dev

Notes

• On MacOS, if you encounter damage or permission issues, run sudo xattr -d com.apple.quarantine /Applications/HyperChat.app
• MacOS nvm users should manually input PATH echo $PATH, the Windows version of nvm seems to work directly.

Telegram

HyperChat user discussion

Calling shell mcp

One-click webpage creation and publish to (cloudflare)

Calling Google Search, asking what TGA Game of the Year is

Organizing Zhihu hot searches

Helping you open a webpage, analyze the results, and write to a file

Opening Baidu and taking a screenshot

Scheduled task list

MCP market (experimental)

Installing MCP interface from the market (experimental)

Installing third-party MCP (supports any MCP)

Installing third-party MCP interface

MCP list (can be dynamically selected)

Rendering HTML, supports Artifacts, SVG, HTML rendering,

Interface 1

Interface 2

Interface 3, testing model capability

Knowledge base

Disclaimer

• This project is for educational and communication purposes only. If you use this project for any actions, such as web scraping, it has nothing to do with the developers of this project.