The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill fetches framework definitions from a public GitHub repository. This is a legitimate operation to synchronize testing procedures with the Open Source Accessibility Framework.
[COMMAND_EXECUTION]: The skill coordinates the use of multiple industry-standard accessibility testing tools, including axe-core, pa11y, Lighthouse CI, and markdownlint. These are standard tools for the stated purpose of accessibility auditing.
[INDIRECT_PROMPT_INJECTION]: As an auditing tool, the skill processes untrusted repository content. While this introduces a potential surface for indirect prompt injection, the skill mitigates risk through a structured workflow that requires evidence for test results and maintainer approval for any proposed changes or issue filings. • Ingestion points: Repository source code, documentation files, and configuration manifests. • Boundary markers: The skill's Output contract enforces a specific reporting structure for findings. • Capability inventory: Includes writing to local documentation files and recommended integration with CI/CD workflows. • Sanitization: Relies on specialized sub-skills for parsing and includes human-in-the-loop approval steps for fixes.

osaf-testing