MCP Server for Smithery Integration

This project implements a FastAPI-based MCP (Model Context Protocol) server that integrates the Greptile API with the Smithery Framework, enabling AI agents to comprehend and interact with codebases.

Project Overview

The MCP Server acts as a critical intermediary between Smithery's AI agents and the Greptile API, providing:

• Standardized MCP-compliant endpoints
• Robust rate limiting
• Authentication and security mechanisms
• Error handling and data transformation
• Consistent response formatting

Rate Limiter Implementation

The rate limiter implementation is defined in app/core/rate_limit.py and has the following components:

Core Components

1. RateLimitStorage (ABC): Abstract base class for storage backends

   • Defines interface with increment, reset, and get_count methods
   • Allows for different storage implementations

2. InMemoryRateLimitStorage: Default implementation

   • Stores counters in memory using Python dictionaries
   • Tracks both counts and window expiry times
   • Not suitable for distributed environments

3. RateLimiter: Main service class

   • Configurable via environment variables
   • Provides check_rate_limit and reset_rate_limit methods
   • Returns detailed limit information

4. FastAPI Integration:

   • Middleware implementation (RateLimitMiddleware)
   • Dependency function (rate_limiter())
   • Standard rate limit headers

Configuration Options

# Settings for rate limiting
RATE_LIMIT_ENABLED: bool = True
RATE_LIMIT_REQUESTS: int = 60
RATE_LIMIT_PERIOD_SECONDS: int = 60

Limitations & Future Enhancements

1. Scalability: Current implementation uses in-memory storage, which doesn't scale across multiple instances. Future implementation should use Redis or similar distributed cache.

2. External API Awareness: Should track Greptile API quotas and adapt rate limits accordingly.

3. Per-Service Limits: Implement separate rate limits for different services and endpoints.

4. Advanced Strategies: Consider implementing token bucket or sliding window algorithms for more sophisticated rate limiting.

Integration with Smithery

This MCP server follows the requirements for integration with the Smithery Framework by:

1. Providing standardized endpoints that map to Greptile functionality
2. Implementing proper authentication and security mechanisms
3. Transforming data between Greptile API format and MCP format
4. Abstracting complex logic into a standardized interface

Development

To run the server locally:

# Set up environment
python -m venv venv
source venv/bin/activate
pip install -r requirements.txt

# Configure environment variables
cp .env.example .env
# Edit .env with your Greptile API key

# Run server
uvicorn app.main:app --reload

License

MIT License