regression-review
Pass
Audited by Gen Agent Trust Hub on Sep 12, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill instructions direct the agent to utilize local system utilities, specifically
openssl rand -hex 4oruuidgen, to generate unique random identifiers for report filenames. This is a standard method for ensuring unique file creation within thetmp/reviews/directory.\n- [INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and analyze untrusted external data such as code changes, pull request descriptions, and issue tracker text, which presents a surface for indirect prompt injection.\n - Ingestion points: Data enters the context via the working tree, staged changes, PR descriptions, issue text, and design documents as specified in
SKILL.md.\n - Boundary markers: The skill does not explicitly instruct the agent to use specific delimiters or protective instructions when processing these external inputs.\n
- Capability inventory: The agent is authorized to write markdown files to the local filesystem and execute the aforementioned system commands for filename generation.\n
- Sanitization: The workflow focuses on structural report generation and does not include explicit steps for sanitizing or escaping external text inputs before they are included in the final report.
Audit Metadata