visual-verify
Pass
Audited by Gen Agent Trust Hub on Apr 1, 2026
Risk Level: SAFECOMMAND_EXECUTION
Full Analysis
- [SAFE]: The skill's behavior is consistent with a legitimate developer tool for UI verification. It provides safety guards such as iteration limits and cost ceilings to prevent runaway execution.\n- [COMMAND_EXECUTION]: The skill instructions involve running common, safe development commands like 'npx tsc', 'npm run lint', and 'npm test' as part of a verification pipeline.\n- [DATA_EXFILTRATION]: Automated checks are targeted at 'localhost' URLs for rendering and screenshot analysis. No sensitive file access or unauthorized network exfiltration was detected.\n- [INDIRECT_PROMPT_INJECTION]: Although the skill analyzes screenshots of web pages (which could technically contain malicious visual instructions), this is a known risk surface for AI-driven UI tools and is mitigated here by its focus on structured layout and CSS verification tasks.
Audit Metadata