cherry-pick-port
Pass
Audited by Gen Agent Trust Hub on Jun 22, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill provides a high-rigor framework for repository interactions. It enforces mandatory pre-action inspections, classifies risk levels (R1-R4), and requires explicit confirmation for mutations. Malicious patterns such as unauthorized network access, secret exposure, or obfuscation are absent.
- [PROMPT_INJECTION]: Analysis of indirect prompt injection vectors shows the skill processes untrusted data from the git repository, such as commit messages and diffs. The risk is mitigated by the skill's restrictive tool rules and specialized routing logic for complex or high-risk actions.
- Ingestion points: Commit metadata and diffs processed by git inspection tools in SKILL.md.
- Boundary markers: None explicitly implemented in prompt templates.
- Capability inventory: Local execution of git commands such as cherry-pick, patch-id, and diff.
- Sanitization: None explicitly present; behavior is constrained by rigid decision rules and local-only execution boundaries.
Audit Metadata