frontend-visual-qa
Pass
Audited by Gen Agent Trust Hub on Jul 24, 2026
Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill utilizes bundled Node.js scripts (visual_layout_audit.mjs and silent_degradation_probe.mjs) to automate layout and font auditing. These scripts are executed locally within the user's project environment.
- [PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection because it ingests and processes content from untrusted web pages during audits. Maliciously crafted web content could theoretically attempt to subvert the agent's instructions.
- Ingestion points: DOM content and rendered text from user-supplied URLs and local HTML files (referenced in SKILL.md and visual_layout_audit.mjs).
- Boundary markers: The skill implements a 'workflow contract' and specific audit profiles to scope the agent's tasks.
- Capability inventory: The skill can drive browser instances and execute local scripts to analyze DOM geometry and visual artifacts.
- Sanitization: Content extraction is performed via standard DOM methods without explicit filtering for adversarial instructions.
Audit Metadata