leverage-and-response-design
Pass
Audited by Gen Agent Trust Hub on May 19, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill's instructions and metadata focus on educational systems thinking frameworks and contain no evidence of malicious behavior, data exfiltration, or obfuscation.
- [INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted user text for analysis, creating a potential attack surface, but no high-risk capabilities were found. Ingestion points: Data enters the agent context through the systems_analysis and proposed_action fields in SKILL.md. Boundary markers: The prompt does not utilize delimiters or specific instructions for the agent to ignore embedded commands. Capability inventory: No specific tools, shell commands, or scripts are provided or utilized by the skill. Sanitization: No input validation or sanitization is performed on the provided analysis text.
Audit Metadata