The Agent Skills Directory

Data Exposure & Exfiltration (LOW): The skill performs network requests to rest.uniprot.org. This domain is not on the trusted whitelist. While appropriate for the skill's purpose, it represents an external communication channel.\n- Indirect Prompt Injection (LOW): The skill retrieves data from an external API, creating a surface for indirect prompt injection.\n
Ingestion points: scripts/uniprot_client.py (via requests calls in search_proteins, get_protein, map_ids, and stream_results).\n
Boundary markers: Absent.\n
Capability inventory: scripts/uniprot_client.py contains only requests for data transfer; no subprocess, exec, eval, or file-write calls were detected.\n
Sanitization: Absent.

uniprot-database