bui21x/huggingface-mcp-server
If you are the rightful owner of huggingface-mcp-server and would like to certify it and/or have it hosted online, please leave a comment on the right or send an email to henry@mcphub.com.
A Model Context Protocol (MCP) server that provides HuggingFace model integration for AI agents.
The HuggingFace MCP Server is designed to facilitate seamless integration of HuggingFace models into AI agent workflows. It supports model inference with GPU acceleration, ensuring efficient processing of tasks such as text generation, classification, and more. The server is equipped with model caching capabilities to enhance performance by reducing redundant loading times. Users can configure custom parameters to tailor the model's behavior to specific needs. Additionally, the server includes health monitoring features, providing real-time insights into GPU status and overall server health. This makes it a robust solution for deploying AI models in production environments, where reliability and performance are critical.
Features
- Model inference with GPU support
- Model caching for improved performance
- Multiple task support
- Custom parameter configuration
- Health monitoring with GPU status
Usages
local integration with stdio
python mcp.run(transport='stdio') # Tools defined via @mcp.tool() decorator
remote integration with sse
python mcp.run(transport='sse', host="0.0.0.0", port=8000) # Specify SSE endpoint
remote integration with streamable http
yaml
paths:
/mcp:
post:
x-ms-agentic-protocol: mcp-streamable-1.0 # Copilot Studio integrationplatform integration with github
{"command": "docker", "args": ["run", "-e", "GITHUB_PERSONAL_ACCESS_TOKEN", "ghcr.io/github/github-mcp-server"]}development framework with fastmcp
python
from mcp.server import FastMCP
app = FastMCP('demo')
@app.tool()
async def query(): ...