blind-red-team
Pass
Audited by Gen Agent Trust Hub on Aug 7, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill consists of pedagogical instructions for performing red-teaming and devil's advocacy based on historical and psychological frameworks (e.g., Ipcha Mistabra). It is entirely text-based and does not interact with the system or network.
- [NO_CODE]: No scripts, binaries, or automated tool configurations are included. The skill serves as a reference for how an agent or user should approach dissent-driven review tasks.
- [PROMPT_INJECTION]: No malicious instructions designed to override safety filters, bypass constraints, or extract system prompts were detected. The instructions encourage critical thinking within the scope of a requested red-teaming task.
- [EXTERNAL_DOWNLOADS]: No URLs, external package dependencies, or remote resource fetches are defined in the skill.
Audit Metadata