aria-text

Pass

Audited by Gen Agent Trust Hub on Jul 31, 2026

Risk Level: SAFE
Full Analysis
  • [SAFE]: The skill is entirely composed of documentation and instructional markdown. It does not include any scripts, binaries, or executable code, minimizing the risk of direct malicious behavior.
  • [SAFE]: All external URLs point to legitimate, well-known resources for web development and accessibility standards, including W3C, MDN, and Frontend Checklist.
  • [PROMPT_INJECTION]: The skill defines a surface for indirect prompt injection because it instructs the agent to analyze external HTML content. However, the risk is negligible as the skill lacks high-risk capabilities such as network exfiltration or file system manipulation. Ingestion points: Reviewing 'rendered HTML' and 'design-system patterns' (SKILL.md). Boundary markers: None identified. Capability inventory: No tools for network or system modification are requested or used. Sanitization: Not applicable for this instructional-only skill.
Audit Metadata
Risk Level
SAFE
Analyzed
Jul 31, 2026, 05:58 PM
Security Audit — agent-trust-hub — aria-text