design-review
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 data which could contain instructions intended to influence the agent's behavior.
- Ingestion points: The skill reads external design briefs (
.design/<feature-slug>/DESIGN_BRIEF.mdor project rootDESIGN_BRIEF.md), project source code components, and visual data from browser screenshots of a running application. - Boundary markers: The instructions do not define explicit delimiters or 'ignore' instructions for the content of the design brief or the application being reviewed.
- Capability inventory: The skill utilizes browser control tools (
browser_navigate,browser_resize,browser_take_screenshot), file system read access for source code, and file system write access for generating theDESIGN_REVIEW.mdreport. - Sanitization: No sanitization, filtering, or validation of the text within the design briefs or the rendered application content is specified before processing.
Audit Metadata