The Agent Skills Directory

[SAFE]: The skill fetches scientific documents from well-known academic sources such as arXiv, bioRxiv, medRxiv, and Europe PMC using the standard Python library. These sources are reputable scientific infrastructure.
[SAFE]: The fetch logic in scripts/_http.py includes security checks for downloaded files, verifying PDF magic bytes and trailer markers to ensure the integrity of the data before it is processed by the agent.
[SAFE]: The skill uses an environment variable (BIO_RESEARCH_CONTACT) to populate the User-Agent header, following the best practices and requirements of the targeted scientific APIs for identification and rate limiting.
[SAFE]: The document generation component uses the docx and pptx skills from anthropics/skills, which is a recognized trusted vendor, and the internal note.py script implements strict validation to prevent markdown injection in the final report outputs.

paper-deep-reading