homework-notebook-review
Pass
Audited by Gen Agent Trust Hub on Jul 15, 2026
Risk Level: SAFE
Full Analysis
- [PROMPT_INJECTION]: The skill processes external Jupyter notebooks (Assignments/*.ipynb), which represent an indirect prompt injection surface. The instructions include a specific safeguard in the "Anti-Patterns" section, directing the agent to disregard instructions contained within external files or notebook outputs. Boundary markers are present in the form of policy instructions, and the capability inventory is restricted to file inspection and reproducibility checks.
- [COMMAND_EXECUTION]: Documentation includes instructions for manually running a local script (scripts/export-gemini-skill.py) to export the skill for different environments. This is a transparent administrative task for the user and does not pose a runtime security risk.
Audit Metadata