lyt-product-image-patterns
Pass
Audited by Gen Agent Trust Hub on Jul 9, 2026
Risk Level: SAFE
Full Analysis
- [PROMPT_INJECTION]: No attempts to override system instructions or bypass safety filters were found. The skill includes clear 'Cause Guardrails' and 'Forbidden Items' to ensure appropriate behavior.
- [DATA_EXPOSURE]: While the skill accesses user-provided image folders and external links for analysis, there is no evidence of unauthorized sensitive data access or exfiltration patterns.
- [REMOTE_CODE_EXECUTION]: The skill does not perform any remote code downloads or dynamic execution. All logic is contained within the prompt instructions and a static knowledge base.
- [INDIRECT_PROMPT_INJECTION]: The skill processes external data (competitor links, image samples). The instructions mandate rigorous data cleaning and factual summarization, mitigating the risk of instructions embedded in external content influencing the agent's core logic.
Audit Metadata