skills/nmnmcc/skills/debate/Gen Agent Trust Hub

debate

Pass

Audited by Gen Agent Trust Hub on Aug 3, 2026

Risk Level: SAFE
Full Analysis
  • [SAFE]: The skill provides instructions for a deliberation process between two AI agents. It does not contain executable code, external network requests, or attempts to bypass safety filters.
  • [PROMPT_INJECTION]: The instructions reinforce safety guidelines, explicitly stating that agent agreement does not grant permission to act beyond the user's original authority. No jailbreak or override patterns were detected.
  • [DATA_EXFILTRATION]: No sensitive data access or external network transmission mechanisms were identified. The subagent is explicitly restricted to read-only checks.
  • [COMMAND_EXECUTION]: The skill does not invoke any shell commands or dynamic execution patterns. It relies on other existing skills (study, just-do-it) for its functional tasks.
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 3, 2026, 06:47 AM
Security Audit — agent-trust-hub — debate