Hugging Face Official MCP Server

**This MCP server is not part of the Databricks Services, and its use is subject to the source's license terms. The MCP server and the information below are reproduced here for your convenience from publicly available listings. Any issues should be submitted to the MCP provider.**
___

**Overview**

The Hugging Face MCP Server connects AI assistants directly to the Hugging Face Hub and thousands of community-built Gradio AI applications. Agents can search models, datasets, Spaces, and papers, query documentation with natural language, and invoke tools from any MCP-compatible Gradio app — all from your editor, chat, or CLI.

What makes this server unique is its extensibility: authenticated users can add any MCP-compatible Gradio Space as additional tools via a [web settings page](https://huggingface.co/settings/mcp). Each Gradio app's functions become callable tools with typed arguments and descriptions, so the server dynamically grows its tool set based on the community's AI applications — image generation, audio transcription, text analysis, and more.

Anonymous users get a standard tool set including Hub search and an image generator. The server supports Streamable HTTP transport and is designed for stateless operation. Over 160 different MCP clients connected in its first week.

Connect at `https://huggingface.co/mcp`.

**Tools**

- **Hub search** — Search for models, datasets, Spaces, and papers on the Hugging Face Hub with metadata, links, and context
- `hf_doc_search` — Search Hugging Face documentation with natural language queries
- `hf_doc_fetch` — Fetch full documentation content for a specific page
- **Gradio Space tools** — Dynamically exposed tools from any MCP-compatible Gradio app you enable — each app's functions become callable tools with typed arguments

Extend the tool set by adding community Gradio Spaces from the [MCP settings page](https://huggingface.co/settings/mcp). Append `?no_image_content=true` to the URL to strip image content blocks from responses.

**Use Cases**

- **Model discovery:** Search for quantized variants, compare architectures, and find models by task or framework.
- **Dataset exploration:** Find datasets for training, evaluation, or fine-tuning workflows.
- **AI application access:** Invoke community Gradio apps (image generation, transcription, text analysis) directly from your IDE without leaving your workflow.
- **Documentation lookup:** Query Hugging Face library docs, API references, and tutorials.
- **Paper research:** Search for and explore ML papers hosted on the Hub.

**Learn More:** [github.com/huggingface/hf-mcp-server](https://github.com/huggingface/hf-mcp-server)