delegated-review-loop
Pass
Audited by Gen Agent Trust Hub on Aug 20, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill implements a sophisticated multi-agent orchestration pattern (Scout, Worker, Reviewer roles) using a strict isolation model.
- [SAFE]: Data handling is restricted to a per-session directory (
.agents/agents-community/<session-id>/), following the principle of least privilege by passing file paths rather than inlining content. - [SAFE]: Metadata and instructional content (SKILL.md) correctly describe the tool's purpose without attempting to override system safety filters or engage in prompt injection.
- [SAFE]: Reference files (
session_protocol.md,reviewer_panel.md,worker_prompt.md,loop_rules.md) provide a robust framework for agent communication and error handling, including automated cleanup procedures.
Audit Metadata