manipulation

Pass

Audited by Gen Agent Trust Hub on Jun 17, 2026

Risk Level: SAFE
Full Analysis
  • [SAFE]: The skill provides documentation and Python code snippets for manipulating biological data using the AnnData library.
  • [EXTERNAL_DOWNLOADS]: The skill references standard, well-known Python packages including anndata, numpy, pandas, scipy, and scikit-learn.
  • [DATA_EXFILTRATION]: No network access or sensitive file system access was detected. The code examples read from local files named metadata.csv and additional_metadata.csv, which are standard names for data science workflows.
  • [COMMAND_EXECUTION]: No shell command execution or privilege escalation patterns were found.
  • [PROMPT_INJECTION]: No attempts to bypass safety filters or override agent instructions were detected.
Audit Metadata
Risk Level
SAFE
Analyzed
Jun 17, 2026, 05:18 AM
Security Audit — agent-trust-hub — manipulation