mcpbeat

Hugging Face MCP Server

co.huggingface/hf-mcp-server
answering

Hugging Face is answering right now. Last checked 14 min ago. It exposes 4 tools.

Connect to Hugging Face Hub and thousands of Gradio AI Applications

Uptime history 42 hours of history
42 hours agonow
100.0%
Uptime 24h
182 of 182 checks
4
Tools
read from the server
133 ms
Response time
average over 24h
API key or OAuth
Access
streamable-http

Connect this server

Endpoint below is the one we actually reach during checks — not the one copied from a README. Last verified 14 min ago.

run in your terminal
claude mcp add hf-mcp-server --transport http https://huggingface.co/mcp?login
~/Library/Application Support/Claude/claude_desktop_config.json
{
  "mcpServers": {
    "hf-mcp-server": {
      "url": "https://huggingface.co/mcp?login"
    }
  }
}
~/.codex/config.toml
[mcp_servers.hf-mcp-server]
url = "https://huggingface.co/mcp?login"
.cursor/mcp.json
{
  "mcpServers": {
    "hf-mcp-server": {
      "url": "https://huggingface.co/mcp?login"
    }
  }
}
.vscode/mcp.json
{
  "mcpServers": {
    "hf-mcp-server": {
      "url": "https://huggingface.co/mcp?login"
    }
  }
}

This endpoint answered with an authorization challenge — the server is running, but you need an API key or OAuth to use it.

Available tools 4

Read directly from the server with tools/list, grouped by what they act on. If a tool disappears, we record the date.

hub
hub_repo_details
Get details for one or more Hugging Face repos (model, dataset, or space). Auto-detects type unless specified. For datasets, use operations: overview, dataset_structure, dataset_preview. Use dataset_structure first to discover configs, splits, sizes, and schema. Use dataset_preview only when config and split are known, unless the dataset has a single config/split.
hub_repo_search
Search Hugging Face repositories with a shared query interface. You can target models, datasets, spaces, or aggregate across multiple repo types in one call. Include links to repositories in your response.
hf
hf_fs
Use to access the Hugging Face Hub. Navigate resources with ls, cat, find, stat, and search over hf:// URIs. Roots: hf://models, hf://datasets, hf://spaces, hf://buckets, hf://collections, hf://papers, hf://docs. For papers, ls hf://papers/ARXIV_ID to discover related resources; cat hf://papers/ARXIV_ID/paper.md or metadata.json. Documentation paths include the current version from each product's llms.txt manifest. Grammar; each token below is one args array element: ls URI [(-R|-r|-lR|-laR|--recursive)] [(-l|-a|-la|-al|--long)] [--glob GLOB] [(-type|--type|--entry-type) TYPE] [--sort SORT] [(-limit|--limit) N] cat URI [RELATIVE_PATH] [(-offset|--offset) N] [(-max-bytes|--max-bytes) N] stat URI [RELATIVE_PATH] find URI [(-R|-r|--recursive)] [(-name|--name|--glob) GLOB] [(-path|--path) GLOB] [(-type|--type|--entry-type) TYPE] [(-limit|--limit) N] search URI [QUERY...] [(-type|--type|--entry-type) TYPE] [--sort SORT] [--tag TAG] [--kind mcp] [(-limit|--limit) N] TYPE = file|dir|repo|bucket|collection|paper|link. Type aliases: f=file, d=dir, l=link, model|dataset|space=repo. SORT = createdAt|downloads|likes|lastModified|likes30d|trendingScore|mainSize|id|trending|upvotes. URI uses hf://, a typed shorthand such as models/OWNER/REPO, or a canonical https://huggingface.co URL. QUERY and GLOB are each one string token. Search URI: hf://models|datasets|spaces[/OWNER], hf://collections[/OWNER], any hf://docs scope, or exactly hf://papers; not hf://. Repository and collection searches may omit QUERY to browse or filter; documentation and paper searches require it. Search joins multiple positional QUERY tokens with spaces. Cat and stat join one RELATIVE_PATH token to URI. Long-list flags are accepted for compatibility; hf_fs listings are already structured, so they do not alter output. Find is already recursive, so recursive flags are accepted without altering behavior. Space search: hf://spaces uses semantic search; repeat --tag to require tags, or use --kind mcp for --tag mcp-server. hf://spaces/OWNER uses owner-scoped keyword search. Documentation: ls hf://docs for products; search any docs scope; use returned hf:// URIs verbatim. Trending listings: ls hf://models/trending, hf://datasets/trending, or hf://spaces/trending. They return up to 20 entries. Trending paths imply trending order; --sort trending|trendingScore is redundant but valid. Trending papers: ls hf://papers/trending. TYPE filters mixed results; omit it when the URI already fixes the result type. Limits and path-specific behavior are documented at hf://README.md. Omit --limit and --sort unless the request asks for a cap, ordering, or exhaustive results. No pipes, redirects, shell expansion, or multiple commands.
whoami
hf_whoami
Hugging Face tools are being used anonymously and may be rate limited. Call this tool for instructions on joining and authenticating.

Endpoints

URLTransportStateLatencyChecked
https://huggingface.co/mcp?login streamable-http needs key 129 ms 14 min ago
https://huggingface.co/mcp streamable-http answering 135 ms 15 min ago

Hugging Face — questions

Answers built from our own checks of this server.

What can Hugging Face do?
It exposes 4 tools, read directly from the server on our last check. Among them: hf_fs, hf_whoami, hub_repo_details, hub_repo_search. The full list with descriptions is on this page — we take it from the server itself via tools/list, not from a README. How MCP servers expose tools in the first place →
Is Hugging Face working right now?
We send a real MCP handshake every 15 minutes. Over the last 24 hours 182 of 182 checks got a reply (100.0%), average response time 133 ms. The bar chart above shows every period we have measured.
How do I connect Hugging Face?
Copy the ready config from this page — we generate it for Claude Code, Claude Desktop, Codex, Cursor and VS Code, each with the file path that client actually reads. It is a remote server, so there is nothing to install — the client connects to the address. You will need your own key or an OAuth login: it refuses anonymous clients.
Does Hugging Face need an API key?
Yes. Every time we knock, Hugging Face answers with an authorization challenge instead of its tool list — that is how we know it is running and gated rather than broken. Bring your own credentials and it will talk.
How fast is Hugging Face?
It answers our handshake in 133 ms on average, which is faster than 77% of all working MCP servers we measure. That puts it in the quick quarter of the ecosystem. The comparison comes from our own checks across the whole registry, every 15 minutes.