paper-retrieval
Use when an agent needs to search scholarly APIs, normalize paper metadata, deduplicate hits, and return a portable ranked candidate list without relying on ChemQA runtime state.
SKILL.mdskills/paper-retrieval/SKILL.md
GitHubPaper Retrieval
Overview
Search OpenAlex, Semantic Scholar, and Crossref from a self-contained script. The output is a normalized candidate list plus diagnostics and provider health.
When to Use
Use this skill when:
- the task starts from a research question or search query
- an agent needs literature candidates outside the current repo workflow
- normalized DOI/title/year/authorship metadata is needed for later access or parsing steps
Do not use this skill for downloading or parsing documents.
Harbor Trial execution
Harbor copies this allowlisted skill to $HOME/.openclaw/skills/paper-retrieval
inside the Trial container. Run the script directly from /workspace; the
retired benchmark runner and host workspace paths are not available.
python "$HOME/.openclaw/skills/paper-retrieval/scripts/paper_retrieval.py" \
--query "Pt/C HER alkaline electrolyte" \
--output-dir /workspace/outputs/paper-retrieval
Read references/contracts.md for request fields and environment variables.