loop-review-fold

Pass

Audited by Gen Agent Trust Hub on Aug 15, 2026

Risk Level: SAFE
Full Analysis
  • [SAFE]: The skill functions as a state machine or router, delegating tasks to other existing skills (review-change, fold-findings, triage-issue) based on the current state of a PR or feature unit.
  • [SAFE]: It emphasizes data integrity by requiring verification of 'durable evidence' (persisted receipts and logs) rather than relying on chat history or potentially inaccurate internal context.
  • [SAFE]: It enforces strict guardrails, explicitly forbidding the router from implementing its own fixes, merging code, or weakening acceptance criteria.
  • [SAFE]: The skill uses standard development tool patterns (feature unit IDs, PR URLs, SHA hashes) and follows a structured output format for user transparency.
  • [SAFE]: Analysis of external references shows they are internal skill delegations within the agent environment, not remote code execution from untrusted third parties.
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 15, 2026, 05:24 PM
Security Audit — agent-trust-hub — loop-review-fold