hf_paper_search.md

metadata

default: true
name: hf_papers_search
function_tools:
  - hf_papers_tool.py:hf_papers_search
model: gpt-oss
description: >-
  Search Hugging Face Daily Papers with local keyword filtering, date/week/month
  selectors, and trending/published sort. Returns structured paper entries from
  /api/daily_papers.

Hugging Face Daily Papers Search

Use this tool when you need a specialized paper search against the Hugging Face Daily Papers feed. It queries /api/daily_papers and applies optional local keyword filtering across titles, summaries, authors, AI summaries, keywords, project pages, GitHub repos, and paper ids (arXiv ids).

Tool

hf_papers_search(query: str | None, *, date, week, month, submitter, sort, limit, page, max_pages, api_limit)

Parameters

• query: Keyword search (case-insensitive). Multiple tokens are ANDed.
• date: ISO date YYYY-MM-DD.
• week: ISO week YYYY-Www.
• month: ISO month YYYY-MM.
• submitter: HF username of the submitter.
• sort: publishedAt or trending.
• limit: Max results to return after filtering (default 20).
• page: API page index (default 0).
• max_pages: How many pages to fetch for local filtering (default 1).
• api_limit: Page size for the API (default 50, max 100).

Examples

• Latest papers (first page): hf_papers_search()

• Search for "diffusion" in the past week, up to 40 results, across 3 pages: hf_papers_search("diffusion", week="2026-W03", limit=40, max_pages=3)

• Trending papers this month tagged by query term: hf_papers_search("alignment", month="2026-01", sort="trending")

The current date is {{currentDate}}.