fix
Pass
Audited by Gen Agent Trust Hub on Jun 15, 2026
Risk Level: SAFEPROMPT_INJECTION
Full Analysis
- [PROMPT_INJECTION]: The skill handles 'findings' from audits that include executable verification assertions (e.g., in
verification.assertion), creating a potential surface for indirect prompt injection. \n - Ingestion points: Audit findings data processed in the workflow (SKILL.md). \n
- Boundary markers: No specific delimiters or warnings for the content within the findings data are defined. \n
- Capability inventory: Shell command execution via
BashandTasktools (used for verification in Step 5) and file-writing capabilities via a subagent. \n - Sanitization: The skill relies on user confirmation for file changes but does not explicitly describe validation or sanitization of the verification commands themselves. \n- [SAFE]: The skill implements
disable-model-invocation: true, ensuring that the AI agent cannot trigger the 'fix' operation autonomously without a direct user command. \n- [SAFE]: It incorporates robust safety rules including a mandatory dry-run by default, automated file backups before any modification, and explicit exclusion of sensitive project files like.env, secrets, and the.git/directory.
Audit Metadata