aesthetic
Pass
Audited by Gen Agent Trust Hub on Sep 23, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill implements a workflow to capture and analyze screenshots from external websites, which serves as an ingestion point for untrusted data. This creates a surface for indirect prompt injection where malicious instructions embedded in a webpage's design could attempt to influence the agent's behavior. Evidence includes the use of chrome-devtools for capturing screenshots and ai-multimodal for analysis without explicit boundary markers or sanitization mentioned in SKILL.md.
- [COMMAND_EXECUTION]: The reference documentation in references/design-resources.md mentions executing gallery-dl via bash to download images. This introduces a command execution surface that could be leveraged if the agent processes unsanitized inputs from external sources.
Audit Metadata