The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process research data and metadata from the public NCBI GEO repository. Because this data is user-submitted, it serves as a potential vector for indirect prompt injection where malicious instructions could be embedded in fields such as study summaries or sample characteristics.
Ingestion points: Untrusted external data is fetched using Bio.Entrez (E-utilities) and the GEOparse library as described in searching.md and geoparse_usage.md.
Boundary markers: The instructions do not define specific delimiters or defensive prompts to prevent the agent from following instructions found within the downloaded biological metadata.
Capability inventory: The agent has the ability to perform network requests to NCBI, write files to the local system, and execute Python scripts using uv.
Sanitization: The skill does not include steps to sanitize or validate the content of the downloaded metadata before processing.
[EXTERNAL_DOWNLOADS]: The skill facilitates the download of data files from NCBI servers. These operations target well-known scientific infrastructure.
Source Details: The skill fetches data from ftp.ncbi.nlm.nih.gov and eutils.ncbi.nlm.nih.gov. These are official, well-recognized domains managed by the National Center for Biotechnology Information.

alterlab-geo