Hugging Face MCP

The Hugging Face MCP server lets AI assistants search and use models, datasets, and Spaces on the Hugging Face Hub, and call hosted inference. It ships as an official remote endpoint so agents can pull ML resources directly into their workflow.

Hugging Face MCP is a AI & LLM MCP server designed for AI engineers, researchers, and product teams. It enables orchestrating models, embeddings, and vector stores from an MCP client by normalising provider APIs into MCP tools for routing and composition. Key capabilities include multi-provider model access, embeddings and similarity search, vector store crud, and structured tool calls that work cleanly inside an LLM context. It integrates with OpenAI, Anthropic, Google, Mistral, Pinecone, Weaviate, Qdrant, and Chroma, and is best suited for AI engineers, researchers, and product teams who need multi-model routing.