Back to directory

paper2code

A Claude Code agent skill that turns an arXiv paper into a citation-anchored code implementation, flagging every choice the paper leaves unspecified.

paper2code

paper2code is an agent skill that converts a research paper into a working implementation. Given an arXiv URL or ID, it fetches and parses the paper, then generates code where every line references the exact section and equation it implements. Before writing code, it classifies each implementation choice as specified, partially specified or unspecified, and marks unspecified choices in the code with the alternatives that were considered, rather than silently guessing.

It treats appendices, footnotes and figure captions as sources rather than ignoring them, and supports a minimal mode, a full mode that includes the training loop and data pipeline, and an educational mode with extra comments and a walkthrough notebook. It installs into Claude Code and other supported agents via a skills add command.

More in Agent skills

See all 581
  • Provenance gate

    github.com

    A deterministic authorization gateway that sits between an AI agent and its tools, tracing where each parameter of a side-effect call came from, and blocking the call or asking a human when the source does not check out.

  • A markdown framework of system rules that restates Isaac Asimov's Four Laws of Robotics as a set of prioritized loss calculations for large-language-model agents to follow, published as a discovery file for agents and crawlers to read.

  • Vitruvius

    github.com

    An engineering research agent that runs a discover, read, synthesize, verify and review loop over technical questions, reading sources directly rather than inferring from titles, and publishing its own scored verifier benchmark, weaknesses included.