accessibility-excellence
Pass
Audited by Gen Agent Trust Hub on Sep 15, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill operates by ingesting, analyzing, and modifying frontend source code to ensure accessibility compliance. This behavior creates a surface where malicious instructions hidden in code comments or string literals could influence the agent's behavior during the audit process.
- Ingestion points: The skill analyzes the project codebase, component stack, and interactive surfaces (forms, modals, widgets) to measure ARIA and semantic HTML usage (SKILL.md, Step 0).
- Boundary markers: The instructions lack specific delimiters or "ignore embedded instructions" warnings for the code context being processed.
- Capability inventory: The skill is authorized to perform source code modifications, including replacing semantic elements, wiring ARIA attributes, and adjusting CSS color tokens. It also uses automated scanning tools like axe-core and Lighthouse (SKILL.md, Step 3 and references/checkpoints.md).
- Sanitization: There is no evidence of sanitization, escaping, or validation of the ingested code content to prevent injection patterns from being interpreted as instructions.
Audit Metadata