Hugging Face Hub Search

Provides semantic search capabilities for Hugging Face models and datasets using vector embeddings to find resources through natural language descriptions, similarity-based discovery, and trending content retrieval with detailed metadata extraction.

This MCP server provides AI-powered semantic search capabilities for Hugging Face models and datasets, built by Daniel van Strien from Hugging Face using a custom search API hosted on Hugging Face Spaces. It offers tools for semantic similarity search that goes beyond keyword matching to find models and datasets based on natural language descriptions, similarity-based discovery to find related resources, trending content retrieval with filtering options, and detailed metadata extraction including safetensors parsing for model architecture analysis and README card downloads. The implementation uses vector embeddings for intelligent search rather than simple text matching, supports parameter count filtering for models, and provides comprehensive filtering options by likes, downloads, and other metrics, making it valuable for researchers and developers who need to discover relevant ML resources through natural language queries or find alternatives to existing models and datasets.