tooluniverse-sequence-analysis

Pass

Audited by Gen Agent Trust Hub on Mar 30, 2026

Risk Level: SAFE
Full Analysis
  • [SAFE]: The skill serves a clear, legitimate purpose for bioinformatics research. It facilitates communication with well-known, trusted scientific repositories (NCBI, Ensembl, UniProt).
  • [DATA_EXFILTRATION]: No evidence of unauthorized data exfiltration. Network operations are restricted to public biological database APIs. No sensitive local file access (e.g., .ssh, .aws) is requested.
  • [REMOTE_CODE_EXECUTION]: The skill defines tool interfaces but does not download or execute remote scripts or unverified binary code.
  • [PROMPT_INJECTION]: The skill processes external data from biological databases, which technically provides a surface for indirect prompt injection. However, the risk is negligible as the skill lacks high-risk write or execution capabilities that could be exploited.
  • Ingestion points: Data retrieved from NCBI, Ensembl, UniProt, and PubMed APIs (SKILL.md).
  • Boundary markers: Not explicitly defined in the retrieval steps.
  • Capability inventory: No subprocess calls, file-write operations, or dynamic code execution capabilities detected.
  • Sanitization: Standard tool response handling is assumed.
Audit Metadata
Risk Level
SAFE
Analyzed
Mar 30, 2026, 11:09 AM