MCP Server for Replicate

A FastMCP server implementation for interfacing with Replicate's API. This server provides tools for accessing various AI models hosted on Replicate through a standardized interface.

Current Status: Early Alpha

This project is in early alpha development. Features and APIs may change significantly.

Currently Supported

• Image generation models with:
• Model schema inspection
• Image generation with customizable parameters
• Output resizing and optimization

Roadmap

Planned Features

1. Text Generation
2. Support for text completion models
3. Chat model integration

4. Streaming support for real-time responses

5. Video Generation

6. Support for video generation models
7. Video output handling and optimization

8. Progress tracking for long-running generations

9. Additional Features

10. Model version management
11. Better error handling and retries
12. Caching for frequently used models
13. Rate limiting and queue management

Setup

1. Install dependencies:

pip install -r requirements.txt

1. Set up your Replicate API token in .env:

REPLICATE_API_TOKEN=your_token_here

1. Run the server:

fastmcp dev server.py