character-animation-qa
Pass
Audited by Gen Agent Trust Hub on Sep 15, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted external content by navigating to a user-provided
previewUrlusing Playwright and capturing screenshots and console logs for agent review. This presents a surface where malicious instructions embedded in the target web page could influence agent behavior. - Ingestion points: The
previewUrlparameter used in the Playwrightpage.goto()command. - Boundary markers: None identified in the provided instructions to delimit external content from system instructions.
- Capability inventory: Playwright browser automation (navigation, screenshots), FFmpeg/ffprobe for media metadata analysis and frame sampling.
- Sanitization: No explicit sanitization or validation of the content retrieved from the URL before visual review by the agent.
Audit Metadata