review-tier

Pass

Audited by Gen Agent Trust Hub on Aug 17, 2026

Risk Level: SAFECOMMAND_EXECUTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill instructs the agent to execute project build, test, and static analysis tasks to verify a branch before starting a review. It provides specific guidance on reliably capturing shell exit codes (e.g., cmd > log; echo $?) within command chains to ensure accurate gatekeeping.
  • [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted code diffs to classify review tiers, but it includes defensive instructions that mitigate potential injection risks.
  • Ingestion points: Changed file paths, symbols, and code diffs processed in Step 1.
  • Boundary markers: Not explicitly specified for the content within the diffs.
  • Capability inventory: Execution of local shell commands for build and test tasks.
  • Sanitization: Not specified; however, the skill explicitly mandates using deterministic patterns and paths rather than model judgment to decide whether to skip a review, which prevents instructions embedded in the code from bypassing security checks.
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 17, 2026, 08:21 AM
Security Audit — agent-trust-hub — review-tier