The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill is designed to interact with approximately 40 bioinformatics web services and databases (e.g., UniProt, KEGG, NCBI, ChEBI, ChEMBL) to retrieve biological information. Use of these well-known scientific services is documented as the primary function of the skill.
[DATA_EXFILTRATION]: The skill sends user-provided biological identifiers and sequences to external APIs for processing. Notably, the NCBI BLAST service requires a user's email address as a parameter to track job status, which is a standard requirement for that specific service.
[INDIRECT_PROMPT_INJECTION]: The skill processes data returned from external APIs (such as protein sequences in FASTA format or annotations in XML/JSON) and reads identifiers from local input files. While these are ingestion points for untrusted data, the skill is focused on structured data manipulation for research purposes, involving a negligible risk of malicious influence on the agent's behavior.
[COMMAND_EXECUTION]: The skill includes executable Python scripts that perform network requests and file I/O operations (reading ID lists, writing CSV/SIF results). The logic within these scripts is transparent and limited to the stated purpose of bioinformatics data analysis.

bioservices