software-accessibility

Pass

Audited by Gen Agent Trust Hub on Oct 6, 2026

Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
  • [EXTERNAL_DOWNLOADS]: The script scripts/run_axe.sh fetches the @axe-core/cli automation tool from the public npm registry via npx. This is a standard engineering practice for accessibility testing.
  • [COMMAND_EXECUTION]: The skill provides automated scripts (scripts/run_axe.sh and scripts/check_a11y_baseline.py) to execute accessibility scans and validate compliance claims.
  • [INDIRECT_PROMPT_INJECTION]: The skill is designed to process external data, specifically accessibility reports and user-provided JSON claims files.
  • Ingestion points: scripts/check_a11y_baseline.py accepts a user-provided JSON file via the --claims argument; scripts/run_axe.sh processes the output of the axe-core scanner.
  • Boundary markers: None explicitly defined in the prompts, though the scripts expect structured JSON input.
  • Capability inventory: The skill has the capability to execute shell commands (npx) and read local files.
  • Sanitization: scripts/check_a11y_baseline.py implements schema validation, type checking, and error handling for the ingested JSON data.
  • [SAFE]: All external references in data/sources.json and SKILL.md target official standards bodies (W3C), government regulators (FTC, DOJ), or well-known industry documentation sites (WebAIM, The A11Y Project).
Audit Metadata
Risk Level
SAFE
Analyzed
Oct 6, 2026, 02:03 PM