credit-note-fixer
Pass
Audited by Gen Agent Trust Hub on Apr 25, 2026
Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
- Command Execution: The workflow requires running a local shell script (
sh tests/test_credit_note.sh). While executing scripts is a standard part of verification in development, it is important to ensure that the content of the script being run is trusted and understood.- Indirect Prompt Injection Surface: The skill instructs the agent to readrepo/task.mdto guide its task. This creates a surface where instructions embedded in project files could influence agent behavior. (1) Ingestion point:repo/task.md. (2) Boundary markers: Not explicitly defined. (3) Capability inventory: File modification viaapply_patchand shell execution viash. (4) Sanitization: No specific sanitization steps are defined for the input from the task file.
Audit Metadata