codepatrol-autofix
Pass
Audited by Gen Agent Trust Hub on Nov 2, 2026
Risk Level: SAFEPROMPT_INJECTIONINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [PROMPT_INJECTION]: The skill explicitly instructs the agent to bypass standard user confirmation steps defined in the software-factory-mode-2026aki development workflow. It directs the agent to proceed autonomously with sensitive operations such as git push, force push, and pull request creation without waiting for approval.
- [INDIRECT_PROMPT_INJECTION]: The skill processes security vulnerability reports and triage pages to determine necessary code changes, which creates a risk of malicious instructions within the reports influencing the agent.
- Ingestion points: Reads security reports and triage data via the cosense skill.
- Boundary markers: Absent; there are no clear delimiters or instructions to ignore embedded commands within the ingested data.
- Capability inventory: Includes git commit, push, force push, and pull request management.
- Sanitization: Absent; the skill does not specify any validation or filtering of the report content before passing it to the code generation tool.
- [COMMAND_EXECUTION]: The skill performs autonomous repository operations, including branch management, rebasing, and force-pushing to remote branches based on its own planning.
Audit Metadata