The Agent Skills Directory

[PROMPT_INJECTION]: The instructions provide a standard operational workflow for accessibility audits and do not contain any instructions intended to override agent behavior or bypass safety guardrails.
[DATA_EXFILTRATION]: No sensitive file paths, hardcoded credentials, or network operations to unknown domains were found. The skill does not attempt to access or export user data.
[REMOTE_CODE_EXECUTION]: The skill mentions standard, reputable accessibility tools such as axe-core, pa11y, and Lighthouse. It provides examples using npx for installation and execution, which is standard practice for developer tools.
[COMMAND_EXECUTION]: Command-line examples are restricted to well-known auditing utilities. There are no patterns involving shell piping from untrusted URLs or privilege escalation.
[DYNAMIC_EXECUTION]: JavaScript snippets are included as integration examples for browser-based testing. These are static templates for the user's local environment and do not execute untrusted logic dynamically.
[OBFUSCATION]: All content is in plain text. No Base64, hex encoding, zero-width characters, or other hidden content techniques were identified.

wcag-audit-patterns