hub-cc-pr-reviewer
Pass
Audited by Gen Agent Trust Hub on Sep 6, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTIONDATA_EXFILTRATION
Full Analysis
- [COMMAND_EXECUTION]: The skill utilizes the GitHub CLI (gh) to retrieve pull request diffs, metadata, and repository protection rules. It also executes a local regression testing script (templates/regression_guard.sh). These actions are consistent with the skill's purpose and stay within the scope of the allowed Bash tool.
- [INDIRECT_PROMPT_INJECTION]: As the skill ingests pull request diffs (which contain untrusted external data), it is technically exposed to indirect prompt injection. However, the implementation includes significant mitigations: it uses a "sanitized capsule" approach for comments, mandates a pre-post audit via a /public-surface-audit command, and explicitly delegates the final merge decision to the user rather than performing it autonomously.
- [DATA_EXFILTRATION]: The skill proactively addresses data leakage risks. It instructs the agent to avoid pasting raw grep results from local memory files into public comments and implements a "fail-closed" security posture: if the privacy audit tool is not configured or unavailable, the review comment will not be posted automatically.
Audit Metadata