What it is

paper-search-pro is a Claude Code skill (it also runs in Codex and any agent that loads SKILL.md) that runs a real multi-source academic literature search and writes a self-contained HTML report.

What problem it solves

Scoping a topic normally means querying several databases by hand and reconciling the results. This skill searches five sources at once — OpenAlex, Semantic Scholar, CrossRef, PubMed and arXiv — classifies relevance with the agent's own parallel SubAgents (no third-party LLM key — your agent is the LLM), and produces a single-file HTML report plus BibTeX/RIS/CSV exports and a PRISMA-S disclosure log. Four depth tiers run from a 5-minute scan to a 3-hour systematic-review prep.

Who uses this workflow

Researchers, graduate students, thesis writers and systematic-review authors who need structured literature discovery.

Example

In Claude Code: Find papers on working memory training in older adults → the skill picks a tier, searches the five sources, classifies relevance, and opens an HTML report. Live demo (real generated report): https://o0000-code.github.io/paper-search-pro/

Notes

• Open source, Apache-2.0. No third-party LLM key — relevance classification runs on the agent itself. For the scholarly APIs, the only required setup is a free OpenAlex API key and an NCBI email (~15 min); Semantic Scholar and CrossRef are optional free add-ons. See references/setup.md.
• Added under Data & Analysis in alphabetical order as an external-repo entry, consistent with existing entries like deep-research and recursive-research, since this is a full multi-file project rather than a single SKILL.md file.