Access and Summarize AI Research Papers

Interacts with Hugging Face Paper pages (hf.co/papers, built on arXiv for AI/CS research), where authors can claim their paper, link associated models, datasets, or Spaces, link a GitHub repo or project page, and link an organization - and where any HF paper or arXiv URL mentioned in a model, dataset, or Space README triggers automatic indexing, though only papers submitted within 14 days of their arXiv publication date can appear on the Daily Papers feed. It first parses a paper ID out of whatever the user provides - a Hugging Face papers URL, its .md variant, an arXiv abs or pdf URL, or a bare arXiv ID - then fetches content as markdown - either via the dedicated .md URL suffix or by sending an Accept: text/markdown header to the normal paper page URL, falling back to the arXiv HTML if no Hugging Face rendering exists - or as structured JSON metadata (authors, media, abstract and AI-generated summary, linked project or GitHub, organization, and upvote count).

curl -s "https://huggingface.co/papers/{PAPER_ID}.md"

Its full endpoint set covers finding models, datasets, or Spaces linked to a paper via an arxiv: filter, claiming authorship (POST /api/settings/papers/claim), fetching the Daily Papers feed with date, week, month, or submitter filters (GET /api/daily_papers), listing papers by cursor (GET /api/papers), hybrid semantic-plus-full-text search across title, authors, and content (GET /api/papers/search), indexing a new paper by arXiv ID (POST /api/papers/index - re-indexing an already-indexed paper is restricted to its authors), and updating a paper's project page, GitHub repo, or organization (POST /api/papers/{paperId}/links, restricted to the author, Daily Papers submitter, or a papers admin).

Use it when a user shares a Hugging Face paper page URL, its markdown variant, an arXiv URL or bare arXiv ID, or asks to summarize, explain, or analyze an AI research paper.

Input: a paper reference in any supported format, parsed down to a bare arXiv ID. Output: paper content as markdown or structured JSON - authors, summary, linked models, datasets, Spaces, GitHub or project links, and upvotes - or a write confirmation for claim, index, or link-update calls.

Runs entirely against huggingface.co's REST API and paper-page markdown endpoints, falling back to arXiv's own abs or pdf pages when Hugging Face lacks sufficient detail. Read-only endpoints need no authentication; write endpoints - claim, index, update links - require Authorization: Bearer $HF_TOKEN.

Developers and agents that need to fetch, search, or manage Hugging Face paper-page content and metadata programmatically rather than browsing hf.co/papers by hand.

A 404 on either the paper page or its .md/API variant most often means the paper simply isn't indexed on Hugging Face yet, rather than a broken request - if a paper ID still can't be found, the next step is verifying the extracted arXiv ID itself, including any version suffix like v1.