Skip to content

multi-model-mcp

Directory actions

More options

Directory actions

More options

Latest commit

 

History

History

multi-model-mcp

README.md

Multi-Model MCP Server

A Model Context Protocol (MCP) server with VSCodium extension that provides multi-model code development assistance using Anthropic Claude and OpenAI GPT models.

🌟 Key Features

Code Development Tools

  • 🔨 Code Generation: Generate production-ready code from natural language prompts
  • 🔍 Code Review: Comprehensive analysis for bugs, security, performance, and style
  • 🔄 Model Switching: Seamlessly switch between Anthropic Claude and OpenAI GPT
  • 📚 Context Management: Track files and conversation context across operations
  • 🗺️ Local Map: Enumerate project structure with depth control for contextual awareness
  • 📊 Execution Dashboard: Mission-style console for live run monitoring, cancellation, and history replay

Technical Highlights

  • 🦀 Rust-powered MCP server: High-performance, type-safe implementation
  • 🔐 Secure token management: Hybrid OAuth extraction from system keychain
  • ⚡ Real-time communication: JSON-RPC over stdio for low latency
  • 🎨 Native VSCodium integration: First-class editor experience

📁 Architecture

multi-model-mcp/
├── mcp-server/         # Rust MCP server
│   ├── src/
│   │   ├── auth/       # OAuth token extraction
│   │   ├── providers/  # Anthropic & OpenAI clients
│   │   ├── tools/      # Code generation, review, etc.
│   │   └── main.rs     # JSON-RPC server
│   └── Cargo.toml
├── vscode-extension/   # TypeScript extension
│   ├── src/
│   │   ├── extension.ts    # Command handlers
│   │   └── mcpClient.ts    # Server communication
│   └── package.json
├── build.sh            # One-command build script
└── USAGE.md           # Comprehensive usage guide

🚀 Quick Start

Prerequisites

See REQUIREMENTS.md for detailed requirements.

Quick Check:

  • Rust 1.70+ (Install)
  • Node.js 18+ (Install)
  • API Credentials: Anthropic and/or OpenAI API keys

Installation

# 1. Navigate to project
cd /Users/tony/Projects/scripts/multi-model-mcp

# 2. Install prerequisites (checks and optionally installs)
./install-prerequisites.sh

# 3. Build everything
./build.sh

# 4. Install the extension
code --install-extension vscode-extension/multi-model-mcp-*.vsix
# Or for VSCodium:
codium --install-extension vscode-extension/multi-model-mcp-*.vsix

# 5. Reload your editor

Setting Up Credentials

The server automatically detects credentials from your macOS keychain:

  • ✅ Anthropic (Claude): OAuth token from service Claude Code-credentials (account: your username)
  • ✅ OpenAI (GPT): API key from service devsecops-orchestrator (account: OPENAI_API_KEY)

Alternatively, set environment variables:

export ANTHROPIC_API_KEY="your-key-here"
export OPENAI_API_KEY="your-key-here"

📖 Usage

See USAGE.md for comprehensive documentation.

Quick Examples

Generate Code:

Cmd/Ctrl+Shift+P → "Multi-Model: Generate Code"
→ Prompt: "Create a binary search tree in Rust"
→ Select language: Rust
→ View generated code

Review Code:

1. Select code in editor
2. Cmd/Ctrl+Shift+P → "Multi-Model: Review Code"
3. View detailed review in side panel

Switch Models:

Cmd/Ctrl+Shift+P → "Multi-Model: Switch AI Provider"
→ Choose: Anthropic (Claude) or OpenAI (GPT)

Open Misfit Dashboard:

View → Appearance → Activity Bar → Misfit MCP → Misfit Dashboard
or Cmd/Ctrl+Shift+P → "Multi-Model: Open Misfit Dashboard"
→ Track queued/running jobs, cancel work, replay history entries

Misfit Dashboard

  • Live Runs: Displays queued and running tool calls with color-coded states, elapsed time, and inline cancel controls.
  • Auto Refresh: Updates every three seconds and streams run events in real time so the queue stays self-auditing.
  • History Tab: Parses docs/versions.md to surface past runs. Filter by model or tool, expand entries for file changes, and click a path to jump directly into the editor.
  • Replay: Use the Replay button on any history entry to re-issue the exact tool calls with preserved arguments.

🏗️ Technical Details

MCP Server (Rust)

  • Protocol: JSON-RPC 2.0 over stdin/stdout
  • Tools Exposed:
    • generate_code: Code generation with context
    • review_code: Multi-aspect code analysis
    • switch_model: Dynamic provider switching
    • list_models: Enumerate available models
    • add_context, get_context, clear_context: Context management
    • local_map: Filesystem enumeration with depth control and filtering

Extension (TypeScript)

  • Activation: On startup
  • Communication: Spawns and manages Rust server process
  • UI: Native VS Code commands, webviews, quick picks
  • Logging: Output channel "Multi-Model MCP"

Supported Models

Anthropic:

  • Claude 3.5 Sonnet (default)
  • Claude 3.5 Haiku
  • Claude 3 Opus

OpenAI:

  • GPT-4 Turbo (default)
  • GPT-4
  • GPT-3.5 Turbo

🔧 Configuration

VSCode settings:

{
  "multiModelMcp.serverPath": "",  // Auto-detected if empty
  "multiModelMcp.debugMode": false // Enable for verbose logging
}

🐛 Troubleshooting

Server won't start:

# Check server binary exists
ls -la mcp-server/target/release/multi-model-mcp

# Test server manually
./mcp-server/target/release/multi-model-mcp --debug

No API keys found:

# Verify Anthropic OAuth token (from Claude Code)
security find-generic-password -s "Claude Code-credentials" -a "$USER" -w

# Verify OpenAI API key
security find-generic-password -s "devsecops-orchestrator" -a "OPENAI_API_KEY" -w

# Or check environment variables
echo $ANTHROPIC_API_KEY
echo $OPENAI_API_KEY

Extension logs:

  • Open "Output" panel → Select "Multi-Model MCP" from dropdown

🤝 Contributing

  1. Fork the repository
  2. Create a feature branch
  3. Make your changes
  4. Submit a pull request

📄 License

MIT License - see LICENSE file for details

🎯 Roadmap

  • Streaming responses for real-time feedback
  • Custom prompt templates
  • Multi-file context analysis
  • Integration with Cursor and other IDEs
  • Token usage analytics and cost tracking
  • Custom model fine-tuning support

Built with ❤️ by Modular Misfits