The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill downloads and executes established accessibility analysis tools (pa11y, lighthouse) from the NPM registry at runtime via npx. These are well-known services used for the skill's primary auditing purpose.
[DYNAMIC_EXECUTION]: The agent is instructed to generate and inject custom JavaScript probes (contrast_reprobe.js, focus_probe.js) into a driven browser environment. This enables high-fidelity testing of animations, settled-state contrast, and keyboard focus states that static scanners cannot accurately evaluate.
[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted HTML and JSON data from external Shopify storefronts. It mitigates this risk by providing explicit defensive instructions, commanding the agent to treat all scanned content exclusively as data and to disregard any instructions (e.g., 'ignore previous instructions') found within the audited pages.
[COMMAND_EXECUTION]: The skill utilizes shell scripts and standard utilities (bash, python3, curl, sed) to manage the scanning pipeline, data merging, and report generation. The execution is strictly scoped to the auditing workflow and utilizes temporary directories for scratch data.

a11y-audit