adverse-review
Pass
Audited by Gen Agent Trust Hub on May 10, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill orchestrates a multi-perspective code review using a defined set of personas (Auditor, Adversary, Pragmatist). All analysis is performed within the platform's execution environment.
- [SAFE]: Data processing is handled by local Node.js scripts (collect.mjs, combine.mjs, synthesize.mjs) that perform deterministic tasks such as gathering source code and merging results. These scripts do not perform network operations or access sensitive system configuration files.
- [PROMPT_INJECTION]: The skill includes explicit instructions in round1.txt to mitigate indirect prompt injection by directing sub-agents to ignore instructions found within the code being analyzed. This addresses the ingestion of untrusted repository data while minimizing execution risk.
Audit Metadata