global-legal-counsel
Pass
Audited by Gen Agent Trust Hub on May 10, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: No malicious behavior or security vulnerabilities were identified in the skill body or reference files.
- [PROMPT_INJECTION]: The skill uses specific triggers and instructions to shift the AI from a neutral general assistant to a biased legal advocate ('Stand on the side of the client'). While these persona-shaping instructions override default behavior, they are legitimate for the skill's purpose and are strictly limited by safety boundaries that prohibit assisting in illegal activities such as destroying evidence or evading regulations.
- [DATA_EXFILTRATION]: The skill implements a comprehensive 'Matter Intake' process that requests sensitive information including financial flows, legal evidence, and internal documents. These requests are consistent with the role of a legal counsel, and there are no instructions to transmit this information to external or unauthorized servers.
Audit Metadata