bio-structure-annotation

Pass

Audited by Gen Agent Trust Hub on Feb 19, 2026

Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTION
Full Analysis
  • [SAFE] (SAFE): No malicious behavior or security risks were found; the skill is focused on structural biology research.
  • [EXTERNAL_DOWNLOADS] (SAFE): Documentation guides users to download tools and weights from reputable scientific sources (mmseqs.com, zenodo.org).
  • [COMMAND_EXECUTION] (SAFE): The skill describes the execution of established bioinformatics tools for their intended analysis purposes.
  • [INDIRECT_PROMPT_INJECTION] (LOW): The skill ingests untrusted protein sequences. 1. Ingestion point: proteins.faa. 2. Boundary markers: None. 3. Capability inventory: Execution of local binaries (tm-vec, boltz, colabfold, foldseek). 4. Sanitization: Not explicitly described, though tools process structured biological data.
Audit Metadata
Risk Level
SAFE
Analyzed
Feb 19, 2026, 05:19 PM