opposing-counsel-attack-en

Pass

Audited by Gen Agent Trust Hub on Aug 25, 2026

Risk Level: SAFE
Full Analysis
  • [SAFE]: The skill provides a framework for the AI to simulate an 'opposing counsel' role to stress-test legal arguments. It uses instructions to shape the AI's persona and output format.
  • [SAFE]: The allowed-tools field correctly limits the skill to the Read tool, preventing unauthorized tool use or network access.
  • [SAFE]: Data residency is set to local, and the skill explicitly includes a 'human gate' requirement, ensuring that the AI's output is reviewed by a professional before use.
  • [SAFE]: All external references (GitHub links to lawvable/awesome-legal-skills) point to well-known, public repositories for legal AI prompts and are handled as trusted attributions rather than executable code.
  • [SAFE]: No obfuscation, prompt injection, or remote code execution patterns were detected. The workflow instructions focus on logical analysis and procedural verification (e.g., [VERIFY] tags).
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 25, 2026, 11:08 AM
Security Audit — agent-trust-hub — opposing-counsel-attack-en