AI导哈

AI编程工具实验平台

一个用于比较和实验不同AI编程工具实现方式的代码库，支持多厂商项目对比与MCP协议实践。

  • by GeorgiaDimaki-AI
  • Python
  • 本地部署
  • 更新于
在 GitHub 查看 GeorgiaDimaki-AI/mcp_playground

使用说明

MCP Playground

A playground repository for experimenting with different AI coding tools and their implementations of various tasks. This repo serves as a comparison and learning space for understanding how different AI assistants approach the same problems.

Purpose

This repository is organized by AI coding vendor/tool, with each having their own folder containing various projects. The goal is to:

  • Compare different AI coding approaches to similar tasks
  • Learn best practices from different AI tools
  • Build a collection of useful implementations
  • Experiment with new AI coding capabilities

Repository Structure

mcp_playground/
├── README.md (this file)
├── prompts/              # Documented prompts for fair comparisons
├── claude_code/          # Projects built with Claude Code
│   └── game-rules-pdf-server/
├── cursor/               # Projects built with Cursor (future)
├── copilot/             # Projects built with GitHub Copilot (future)
└── ...                  # Other AI tools as needed

Prompts for Fair Comparison

Each project includes a PROMPT.md file documenting the original prompt used to generate it. This enables fair comparisons across different AI coding tools.

See: prompts/README.md for a list of available prompts and how to use them for comparison testing.

Current Projects by Vendor

Claude Code

Located in claude_code/ - See claude_code/README.md for details.

Projects:

  1. game-rules-pdf-server - An MCP server for querying game rules from PDF files

Other Vendors

Additional folders can be added for other AI coding tools:

  • cursor/ - Projects built with Cursor
  • copilot/ - Projects built with GitHub Copilot
  • codeium/ - Projects built with Codeium
  • tabnine/ - Projects built with Tabnine
  • etc.

How to Use This Repo

  1. Browse by vendor: Explore different AI tools' approaches in their respective folders
  2. Compare implementations: Look at how different tools solve similar problems
  3. Try the projects: Each project has its own README with setup instructions
  4. Add your own: Create new vendor folders and projects as you experiment

Contributing

This is a personal playground, but the structure can be adopted by anyone interested in comparing AI coding tools. Feel free to:

  • Add new vendor folders
  • Add new projects within existing vendor folders
  • Document interesting findings or comparisons
  • Experiment with cross-vendor collaborations

Why MCP?

Many projects in this playground focus on the Model Context Protocol (MCP) because it provides:

  • A standardized way to extend AI assistants
  • Practical, real-world use cases
  • Good test cases for comparing AI coding capabilities
  • Useful tools that can be reused across different AI platforms

License

MIT License - See individual projects for specific licenses if different.

Getting Started

  1. Choose a vendor folder (e.g., claude_code/)
  2. Navigate to a project (e.g., game-rules-pdf-server/)
  3. Read the project's README for setup instructions
  4. Try it out and compare with other implementations!

Last Updated: 2025-10-22 Active Vendors: Claude Code Total Projects: 1

相关 MCP