issue-resolve
Pass
Audited by Gen Agent Trust Hub on Sep 20, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted data in the form of bug reports and issue assessments. This ingestion point is a core component of the bug-fixing workflow. The risk is mitigated by a strict "Autonomy envelope" that prevents the agent from taking irreversible or high-risk actions, such as merging code or spending API credits, without explicit human intervention.
- [COMMAND_EXECUTION]: The skill automates several CLI-based development tasks, including git worktree management and GitHub pull request creation. These actions are restricted to non-protected branches and draft PRs, ensuring a human-in-the-loop requirement before any code changes can impact the primary codebase.
- [SAFE]: The skill includes comprehensive operational guardrails that explicitly forbid the agent from claiming unverified fixes, spending financial credits (Veo), or bypassing the human review process for code promotion.
Audit Metadata