roundtable

Pass

Audited by Gen Agent Trust Hub on Aug 21, 2026

Risk Level: SAFE
Full Analysis
  • [SAFE]: No security issues were detected. The skill consists of prompt-based instructions for organizing sub-agent debates and does not contain executable code, remote script downloads, or requests for sensitive data.
  • [INDIRECT_PROMPT_INJECTION]: While the skill processes user-provided motions and evidence packs, which are then analyzed by sub-agents, the risk is minimized as the sub-agents are restricted to a text-based debate format and are not granted access to external tools or file system modifications within the skill's logic.
  • Ingestion points: User-provided "motion" and "evidence pack" in SKILL.md.
  • Boundary markers: Sub-agents are provided with isolated packs of evidence.
  • Capability inventory: No tools or code execution capabilities are defined for the sub-agents.
  • Sanitization: Not applicable as no executable actions are performed.
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 21, 2026, 01:22 AM
Security Audit — agent-trust-hub — roundtable