Skip to content
Kitbay
Skill

literature-search-arxiv

Apache-2.0Updated on Nov 10, 20263,503GitHub starsIncludes scripts

An arXiv skill in Google DeepMind's science-skills collection, which is built mainly for Google Antigravity's Science plugin and can also be added to other agents with npx skills.

What the skill teaches

An arXiv skill in Google DeepMind's science-skills collection, which is built mainly for Google Antigravity's Science plugin and can also be added to other agents with npx skills. Bundled Python scripts, run through uv, search the arXiv API with field prefixes, ID lists, paging and sorting and return JSON metadata, download a paper as PDF or HTML, and fetch the LaTeX source archive. The skill tells the agent to keep to arXiv's limit of one request every three seconds, to show the user arXiv's API terms once and record that notice, to extract archives only into a new folder, and to list the URLs of every paper it used.

Capabilities

  • Search arXiv with field prefixes and boolean queries, returning JSON
  • Fetch papers directly by arXiv ID
  • Download a paper as PDF or HTML
  • Download a paper's LaTeX source archive