full-code-review
Pass
Audited by Gen Agent Trust Hub on Sep 6, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [SAFE]: The skill implements a multi-stage review process with an adversarial verification step to minimize false positives and speculative findings. It proactively protects against sensitive data exposure by redacting secrets and emails from diffs before they are processed by the LLM.
- [COMMAND_EXECUTION]: Utilizes git and gh via the Bash tool to fetch diffs and PR metadata for analysis. This usage is scoped to read-only operations as defined in the skill's frontmatter.
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted PR diffs and commit logs. It mitigates injection risks through explicit boundary markers, automated secret redaction in scripts/full-code-review.js, and an adversarial verification agent that scrutinizes every finding.
Audit Metadata