experience-accessibility-validate
Pass
Audited by Gen Agent Trust Hub on Sep 19, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill is designed to analyze untrusted user-provided source code (LWC HTML, JS, and CSS). This creates an inherent attack surface for indirect prompt injection where instructions embedded in the code could attempt to influence the agent's behavior.
- Ingestion points: User-provided Lightning Web Component files (HTML, JS, CSS) analyzed in the references/reviewers context.
- Boundary markers: The skill contains detailed instructions and a specific framework that constrains the agent to focus exclusively on WCAG criteria, acting as a functional boundary.
- Capability inventory: The skill uses a bundled Python script for calculations and creates a markdown report (
a11y-review.md). - Sanitization: The skill relies on the agent's analysis logic to identify accessibility issues; no specific sanitization of the input code strings is performed prior to processing.
- [COMMAND_EXECUTION]: The skill invokes a local Python script (
scripts/contrast-ratio.py) to compute WCAG relative luminance and contrast ratios. This script is a static, bundled component of the skill and is used for mathematical computation. It uses standard libraries and is executed via the platform's permittedpython3tool as specified in the skill metadata.
Audit Metadata