redteam

Pass

Audited by Gen Agent Trust Hub on May 27, 2026

Risk Level: SAFEPROMPT_INJECTION
Full Analysis
  • [PROMPT_INJECTION]: The skill employs adversarial directives like "Think like an attacker," "break this," and "real attackers don't follow rules." While these are contextually appropriate for the skill's intended purpose of security review, they utilize patterns typically associated with role-play or safety bypass attempts.
  • [PROMPT_INJECTION]: The skill processes untrusted user data via the $ARGUMENTS variable without providing boundary markers or instructions to ignore embedded commands, creating a surface for indirect prompt injection.
  • Ingestion points: The $ARGUMENTS placeholder at the end of SKILL.md.
  • Boundary markers: No delimiters (such as XML tags or triple backticks) are used to encapsulate the user input.
  • Capability inventory: The skill does not define any allowed-tools and lacks instructions for network, file system, or subprocess operations.
  • Sanitization: No evidence of input sanitization or filtering was found.
Audit Metadata
Risk Level
SAFE
Analyzed
May 27, 2026, 02:35 PM
Security Audit — agent-trust-hub — redteam