subagent-driven-development
Pass
Audited by Gen Agent Trust Hub on May 7, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill implements a structured orchestration workflow that delegates tasks to specialized sub-agents with narrow, task-specific contexts, which is a best practice for minimizing context pollution and improving reliability.
- [SAFE]: The prompts for the reviewer sub-agents include explicit defensive instructions such as "Do not trust the report" and "Verify everything independently," which are effective measures against hallucination or intentional sub-agent misbehavior.
- [SAFE]: The workflow requires the use of isolated environments (via git worktrees) and mandates a multi-stage review loop where implementation must be verified against the original specification before proceeding to quality checks.
- [SAFE]: All external tools referenced (e.g., superpowers:using-git-worktrees) appear to be part of the vendor's intended development ecosystem and are used for legitimate software engineering tasks like test execution and branch management.
Audit Metadata