The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process untrusted data in the form of HTML code and external URLs for accessibility auditing. This creates a surface for indirect prompt injection if the target content contains malicious instructions meant to influence the agent. However, the risk is mitigated by the included evaluation suite and schema validation.
Ingestion points: SKILL.md (URL inputs) and evals/accessibility-testing.yaml (HTML fixtures for test cases).
Boundary markers: The evaluation suite (evals/accessibility-testing.yaml) uses strict must_contain and must_not_contain assertions to ensure the agent focuses on accessibility criteria.
Capability inventory: Browser automation via qe-browser, fleet coordination through FleetManager, and CLI execution via aqe and vibium.
Sanitization: Output is strictly validated against the schemas/output.json schema, which constrains the agent's response to specific accessibility metrics and categories.
[COMMAND_EXECUTION]: The skill references platform-specific CLI tools like aqe and vibium for running accessibility audits. These commands are documented for use within a specialized testing environment and are functionally consistent with the skill's primary purpose of automated quality engineering.
[EXTERNAL_DOWNLOADS]: The skill utilizes well-known, industry-standard accessibility testing libraries and engines, including axe-core, pa11y, Lighthouse, and @axe-core/playwright. These dependencies are typical for this domain and do not represent a security risk when used as instructed.

accessibility-testing