generated: '2026-06-20'
method: searched
status: published
source: https://huggingface.co/docs/hub/en/hf-mcp-server
server:
  name: hugging-face
  transport:
  - stdio
  - streamable-http
  url: https://huggingface.co/mcp
  settings: https://huggingface.co/settings/mcp
  source_code: https://github.com/huggingface/hf-mcp-server
  auth: Bearer User Access Token (hf_...) or OAuth. Anonymous access is available for read/search tools;
    authenticated access unlocks jobs and private repos.
notes: Official Hugging Face remote MCP server. Connects MCP-compatible clients (Claude Desktop, Cursor,
  VS Code, Zed, Codex, ChatGPT) to the Hub. Beyond the built-in tools below, users can add MCP-compatible
  Gradio Spaces as additional runtime tools from the settings page.
tools:
- name: Spaces Semantic Search
  description: Find the best AI apps (Spaces) via natural language queries.
- name: Papers Semantic Search
  description: Find ML research papers via natural language queries.
- name: Model Search
  description: Search for ML models with filters for task, library, and more.
- name: Dataset Search
  description: Search for datasets with filters for author, tags, and more.
- name: Documentation Semantic Search
  description: Search the Hugging Face documentation using natural language across all Hugging Face libraries.
- name: Run and Manage Jobs
  description: Run, monitor, and schedule jobs on Hugging Face infrastructure.
- name: Hub Repository Details
  description: Get detailed information about models, datasets, and Spaces, optionally including README
    content.
deployment:
  mode: remote
  endpoint: https://huggingface.co/mcp
  verified: probed
  probe: live
  checked: '2026-08-12'
  source: catalog MCP census