sensory-instructions

Pass

Audited by Gen Agent Trust Hub on Aug 11, 2026

Risk Level: SAFE
Full Analysis
  • [SAFE]: The skill is entirely educational and instructional, focusing on frontend accessibility standards (WCAG) regarding sensory-only instructions. It does not contain executable scripts, shell commands, or network exfiltration patterns. All provided links point to legitimate documentation at frontendchecklist.io.
  • [INDIRECT_PROMPT_INJECTION]: While the skill instructs the agent to process external data (rendered markup and interactive components), it lacks any high-privilege capabilities or tools that could be exploited by malicious inputs within that data.
  • Ingestion points: The skill specifies reviewing 'rendered markup and interactive states'.
  • Boundary markers: Absent; instructions do not define specific delimiters for the analyzed code.
  • Capability inventory: None; the skill does not use subprocesses, file-writing, or network tools. Its output is limited to reporting accessibility violations.
  • Sanitization: Absent; the skill assumes the LLM will identify violations without executing the provided markup.
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 11, 2026, 12:29 AM
Security Audit — agent-trust-hub — sensory-instructions