The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill interacts with reputable academic services to resolve identifiers and download content. It makes network requests to api.unpaywall.org for DOI resolution, arxiv.org (and its API) for preprints, and dl.acm.org as a fallback for ACM Digital Library papers. These are well-known, established services for scholarly research.
[SAFE]: Input validation is rigorously implemented in scripts/resolve_oa.py using regular expressions (DOI_RE, ARXIV_NEW_RE, ARXIV_OLD_RE) to ensure only valid academic identifiers are processed, mitigating injection risks.
[SAFE]: The skill implements a robust politeness contract, including enforced rate limits (3 seconds for arXiv), exponential backoff for 429/503 status codes, and mandatory identification via a User-Agent string containing a contact email.
[SAFE]: Data handling is restricted to transient processing. The script enforces response size limits (MAX_API_BYTES, MAX_PDF_BYTES) and ensures downloaded PDFs are stored in temporary directories or a local cache, explicitly avoiding persistent storage in the repository.
[SAFE]: No obfuscation, hardcoded credentials, or persistence mechanisms were detected. The Python script uses only the standard library, avoiding supply chain risks associated with external packages.

fetch-paper