charles-unslop
Pass
Audited by Gen Agent Trust Hub on Nov 7, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill instructs the agent to audit and modify user-provided UI code and screenshots, creating a surface where malicious instructions embedded in the target UI could potentially influence the agent.
- Ingestion points: The agent ingests project code and screenshots during the 'Audit' phase (SKILL.md, references/signs.md).
- Boundary markers: The skill enforces a structured 'Direction' and 'Audit' workflow which acts as a logical boundary, though no explicit string sanitization is present.
- Capability inventory: The skill is capable of modifying HTML, CSS, and JS files, and it can interact with image generation tools (SKILL.md, references/visuals.md).
- Sanitization: The skill lacks explicit instructions to sanitize or filter text content found within the processed UI elements.
Audit Metadata