Guide · Agent skill
How to use the Hugging Face Paper Lookup agent skill
Updated · written with AI from the project's own documentation and checked against our catalogue data
The huggingface-papers skill lets an AI agent read Hugging Face paper pages and arXiv papers, then pull structured details such as authors, linked models, datasets, Spaces, the GitHub repository and the project page. After reading this guide you will know how to install the skill, trigger it with a natural request, and use the API calls it documents for search, listing and Daily Papers.
Hugging Face Paper Lookuphuggingface/skills · Apache-2.0 Download the skill Quick start
- 1Download and unzip. Download the huggingface-papers ZIP from this page and unzip it so you have a folder containing SKILL.md.
- 2Place the folder. Move it into ~/.claude/skills/ for all projects, or into the project's .claude/skills/ folder for one project.
- 3Or upload in the apps. In the Claude apps, go to Settings → Capabilities → Skills and upload the skill.
- 4Make sure curl is available. The skill's commands use curl, so confirm it is installed in your shell.
- 5Set your token if needed. For search, listing, Daily Papers and claiming, set HF_TOKEN to your Hugging Face API token.
- 6Share a paper. Paste a Hugging Face paper URL, arXiv link or arXiv ID and ask for a summary, explanation or analysis.
What is an agent skill, and what does this one do?
An agent skill is a folder containing a SKILL.md file that tells an AI agent when and how to perform a task. This skill is built around Hugging Face Paper pages (hf.co/papers), a platform layered on top of arXiv for AI and computer science research papers.
When you share a Hugging Face paper page link, an arXiv URL, or an arXiv ID, the agent can fetch the paper content as markdown and read structured metadata. It can then summarize, explain or analyze the paper.
- Fetch paper content as markdown from a Hugging Face paper page or arXiv ID
- Retrieve structured metadata: authors, linked models, datasets and Spaces, GitHub repo and project page
- Search papers, list papers, and browse the Daily Papers feed through the documented API
- Claim authorship of a paper for a Hugging Face user, using a token
Install the skill from the ZIP or the Claude apps
Download the ZIP for the huggingface-papers skill from this page and unzip it. Place the resulting folder in one of these locations, depending on how you want to use it:
- Personal use across projects: unzip into ~/.claude/skills/
- Single project only: unzip into the project's .claude/skills/ folder
- Claude apps: upload the skill under Settings → Capabilities → Skills
Trigger the skill with natural requests
You do not need a special command. The skill is meant to activate when your request involves a paper. The source lists these triggers: sharing a Hugging Face paper page URL, sharing an arXiv URL or ID, or asking the agent to summarize, explain or analyze an AI research paper.
- "Summarize https://huggingface.co/papers/2602.08025 for me."
- "Explain the main idea of arXiv 2602.08025 in plain language."
- "Who wrote this paper and what models are linked to it? arxiv.org/abs/2602.08025"
- "Analyze this AI paper and list its GitHub repo and project page."
Understand the paper ID and the markdown fetch
The skill first converts whatever you provide into an arXiv paper ID. Hugging Face paper URLs, markdown paper URLs, arXiv abstract and PDF URLs, bare IDs and versioned IDs such as 2602.08025v1 are all accepted.
To read the paper as markdown, the agent runs a curl request against the paper ID. The content relies on the HTML version of the paper on arXiv. If that HTML does not exist, the content falls back to the Hugging Face paper page HTML. A 404 means the paper is not yet indexed on hf.co/papers.
curl -s "https://huggingface.co/papers/{PAPER_ID}.md"Use the API for structured metadata and linked resources
For structured details, the skill calls the Hugging Face REST API. The metadata can include author names and usernames for claimed papers, media URLs uploaded at Daily Papers submission, the abstract, an AI-generated summary, the project page, the GitHub repository, and upvote counts.
Linked models, datasets and Spaces are found with filter queries on the arXiv ID.
- Metadata: returns authors, summaries, project page and GitHub repository
- Models, datasets and Spaces linked to the paper are listed by filter
curl -s "https://huggingface.co/api/papers/{PAPER_ID}"Search, list and browse papers with a token
The search, list and Daily Papers endpoints send a Hugging Face token in the Authorization header. The source names this variable HF_TOKEN. Search is hybrid semantic and full-text and covers the paper title, authors and content, with a query limit of 250 characters and up to 120 results.
- Daily Papers: filter by date, ISO week, month or submitter, and sort by publishedAt or trending
- List papers: paginate with a cursor, between 1 and 100 results per page
- Search: the q parameter takes up to 250 characters
curl -s -H "Authorization: Bearer $HF_TOKEN" "https://huggingface.co/api/papers/search?q=vision+language&limit=20"Safety notes: scripts and network access
This skill runs scripts and uses the network. Its commands call huggingface.co, and the claim and index endpoints send POST requests. The claim endpoint requires a Hugging Face token and should only be used by the actual author of the paper.
Check the commands the agent proposes before running them, especially anything with a POST request or your token in the header. Keep HF_TOKEN in your environment rather than pasting it into chats.
- Only the paper's actual author should claim authorship
- Review POST requests before they run
- Verify key claims from summaries against the original paper
What people use it for
- Getting a quick explanation of an AI paper a colleague sent you
- Finding the GitHub repo, project page and linked models for a paper you are reading
- Browsing the Daily Papers feed for a given date, week or month
- Searching papers by topic such as vision language models
- Checking which datasets and Spaces are linked to a paper before trying them out
Frequently asked questions
How do I install the huggingface-papers skill?
Download the ZIP from this page and unzip it. Put the folder in ~/.claude/skills/ for personal use or in a project's .claude/skills/ folder, or upload it in the Claude apps under Settings → Capabilities → Skills.
Is the Hugging Face Paper Lookup skill free?
The skill is listed under the Apache-2.0 licence, as shown on this page. Check the licence file in the repository for the full terms. The skill itself does not state a price, and the API calls depend on a Hugging Face account for authenticated requests.
What do I need to use it?
You need curl available in your shell. Search, listing, Daily Papers and claiming authorship need a Hugging Face API token, set as HF_TOKEN. Basic paper fetching is done with curl requests to huggingface.co.
What kinds of papers does it cover?
The platform focuses on AI and computer science research papers, so papers in other fields are outside its scope. Papers are based on arXiv, so an arXiv ID or URL is the usual input.
Can the skill guarantee accurate citations?
No. Summaries come from the paper content and AI-generated summaries, so verify key claims against the original paper before citing it.
Why does a paper return a 404 or show no metadata?
A 404 means the paper is not yet indexed on hf.co/papers. Not every arXiv paper has an HTML version, in which case the content falls back to the Hugging Face page HTML.
New agent skills, repos and AI tools — every week
The best new skills and open-source projects, the fastest-growing repos, and one free prompt. Five minutes to read. Unsubscribe anytime.
Our verdict
This skill suits students and engineers who want a fast, sourced explanation of an AI paper, and researchers who need linked code, models and Daily Papers in one place. It is a poor fit for citation-grade accuracy, for papers outside AI and computer science, or for anyone wanting to claim authorship for someone else.
Download the skill