red-team-tribunal
Pass
Audited by Gen Agent Trust Hub on Mar 26, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill implements a structured framework for security auditing and code quality validation using a multi-agent consensus mechanism. No malicious behavior or safety violations were identified.
- [COMMAND_EXECUTION]: The skill uses the Bash tool to execute a local Python script (
red-team-tribunal.py) for analyzing target files and diffs. This execution is scoped to the skill's intended purpose and occurs within the local file system. - [PROMPT_INJECTION]: As a code review tool, the skill processes external input (source code, PR diffs) which represents a surface for indirect prompt injection. However, the system's adversarial architecture, specifically the 'Skeptic' agent, is explicitly designed to detect and mitigate logic errors and security flaws in processed data.
Audit Metadata