code-review-in-stack
Pass
Audited by Gen Agent Trust Hub on Sep 16, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill is designed for code review and explicitly forbids editing code or publishing content without human approval.
- [COMMAND_EXECUTION]: The skill uses standard development tools like
gh(GitHub CLI) and local scripts (stack-map.sh) to gather context about pull requests and stacks. These are used for read-only operations and are consistent with the skill's stated purpose. - [DATA_EXPOSURE]: The skill accesses repository metadata (PR IDs, shas, tickets) and local documentation (
docs/agents/issue-tracker.md). This access is scoped to the repository being reviewed and does not involve sensitive credentials or exfiltration to external domains. - [PROMPT_INJECTION]: No instructions were found that attempt to bypass safety filters or override the agent's core identity. The 'Hard rules' section reinforces safety by limiting the agent's autonomy (e.g., 'Edit no code', 'Publish nothing without approval').
- [INDIRECT_PROMPT_INJECTION]: While the skill processes untrusted data (PR bodies and tickets), it does so within a structured review framework. It does not execute code from these sources or use them to dynamically generate commands that could lead to privilege escalation. The risk is low and mitigated by the requirement for human review of all outputs.
Audit Metadata