RaiAnsar/claude_code-coding-mcp

MCP AI Collab - Give AI Assistants Persistent Memory 🧠

Transform stateless AI assistants into intelligent collaborators with persistent, project-aware memory.

🎯 The Problem We Solve

When you use AI assistants like Gemini, Grok, or ChatGPT through Claude Code, they forget everything between calls:

You: "Hey Gemini, help me debug this auth function"
Gemini: "I found the issue on line 42..."

// 5 minutes later...
You: "Gemini, what was that issue we found?"
Gemini: "I don't have any previous context..."  😔

This MCP server fixes that! Each AI now remembers your conversations per project:

You: "Gemini, what was that issue we found?"
Gemini: "We found a race condition in the auth function on line 42. 
         We discussed using a mutex lock to fix it." 🎯

🌟 Key Features

• 🧠 Persistent Memory - Each AI maintains conversation history per project
• 📁 Project Isolation - Separate contexts for different projects automatically
• 🚀 Three Versions - Choose based on your needs (see below)
• 🔒 100% Secure - Your API keys never leave your machine
• ⚡ Fast Performance - Redis caching with PostgreSQL persistence
• 🔧 Easy Setup - One-click installation with interactive menu

📋 Choose Your Version

We offer three versions to match your needs:

See →

🚀 Quick Start

# Clone the repository
git clone https://github.com/RaiAnsar/claude_code-coding-mcp.git
cd claude_code-coding-mcp

# Run interactive setup (recommended)
./one_click_setup.sh

The setup wizard will:

1. Check your system requirements
2. Let you choose a version
3. Guide you through API key configuration
4. Install everything automatically

🔑 Configuration

Getting API Keys (Required)

You'll need at least one API key:

• Gemini: Google AI Studio (Free tier available)
• Grok: X.AI Console (Requires X Premium+)
• ChatGPT: OpenAI Platform (Pay as you go)

Setting API Keys

During setup, you'll be prompted to enter your keys. They're stored in a local .env file:

GEMINI_API_KEY=your-gemini-key-here
GROK_API_KEY=your-grok-key-here
OPENAI_API_KEY=your-openai-key-here

Customizing AI Models

You can customize which models to use for each AI service in your .env file:

# Default models (latest versions)
GEMINI_MODEL=gemini-2.5-pro-preview-06-05
GROK_MODEL=grok-3
OPENAI_MODEL=gpt-4o

# You can change to other available models:
# GEMINI_MODEL=gemini-2.0-flash-001
# OPENAI_MODEL=gpt-4o
# GROK_MODEL=grok-2

Simply edit your .env file to use different models based on your needs (cost, performance, capabilities).

📖 How to Use

After installation and restarting Claude Code:

Basic Commands

# Check if everything is working
Use db_status

# Ask AIs questions (they'll remember context)
Use ask_gemini to explain this authentication flow
Use ask_grok to help optimize this algorithm
Use ask_openai to review our API design

# Check conversation history
Use show_context with ai "gemini"

# Clear memory for fresh start
Use clear_context with ai "all"

Real Example Session

You: Use ask_gemini to analyze the performance bottleneck in our API

Gemini: I can see the main bottleneck is in the database query on line 234...

You: Use ask_gemini to suggest optimization strategies

Gemini: Based on our previous analysis of the bottleneck on line 234, 
        here are three optimization strategies...
        [Gemini remembers the context!]

🏗️ Architecture

graph TD
    A[Claude Code] -->|MCP Protocol| B[MCP AI Collab Server]
    B --> C{Router}
    C --> D[Gemini API]
    C --> E[Grok API]
    C --> F[OpenAI API]
    B --> G[Context Manager]
    G --> H[Redis Cache]
    G --> I[PostgreSQL]
    G --> J[Local JSON]
    
    style A fill:#f9f,stroke:#333,stroke-width:4px
    style B fill:#bbf,stroke:#333,stroke-width:4px
    style G fill:#bfb,stroke:#333,stroke-width:4px

🔒 Security & Privacy

Your API keys are 100% safe:

• ✅ Local Storage Only - Keys are stored in .env file on your machine
• ✅ Never Transmitted - Except to official AI APIs (Google, X.AI, OpenAI)
• ✅ Never Logged - No keys in logs, console output, or error messages
• ✅ Gitignored - .env files are excluded from version control
• ✅ Open Source - Review our code anytime

→

📚 Documentation

• - Get running in 5 minutes
• - Technical deep dive
• - Choose the right version
• - Common issues and solutions
• - Real-world usage patterns

❓ FAQ

Q: Is this trying to steal my API keys?
A: No! Your keys stay on your machine. Check our source code - we're fully open source.

Q: Which version should I use?
A: Start with the Clean version. You can always upgrade to Full later.

Q: Can I use just one AI?
A: Yes! You only need API keys for the AIs you want to use.

Q: Does this work with Claude's /clear command?
A: Yes! When you clear Claude's context, it clears the AI contexts too.

Q: How is this different from using AIs directly?
A: This gives them memory within Claude Code, making them true collaborators.

🤝 Contributing

We welcome contributions! See for guidelines.

📜 License

MIT License - see for details.

🙏 Acknowledgments

Built on top of the MCP protocol by Anthropic.

Ready to give your AI assistants persistent memory? Get started now! 🚀