python-debug-execution
Pass
Audited by Gen Agent Trust Hub on Jun 24, 2026
Risk Level: SAFE
Full Analysis
- [COMMAND_EXECUTION]: The skill provides patterns for executing local Python scripts using the
python3command as part of a debugging workflow. - [DATA_EXFILTRATION]: The recommended diagnostic code uses
os.getcwd()andos.listdir('.')to print the current working directory and its file list. This exposes environment metadata to the agent's context for troubleshooting purposes. - [PROMPT_INJECTION]: The skill implements a pattern that redirects stderr to stdout (
2>&1), ensuring all script output is captured into the agent's context. This creates an ingestion point for data that could potentially contain indirect prompt instructions if the script processes untrusted files.
Audit Metadata