The Agent Skills Directory

[SAFE]: The skill acts as an educational resource and instruction set for accessibility reviews. All external references, such as to W3C standards and frontendchecklist.io, are to reputable and well-known industry sources.
[INDIRECT_PROMPT_INJECTION]: The skill possesses an attack surface for indirect prompt injection as it is designed to analyze external, potentially untrusted markup.
Ingestion points: The agent is instructed in SKILL.md to review rendered HTML, interactive components, and design-system patterns.
Boundary markers: SKILL.md does not specify the use of delimiters or provide instructions to disregard embedded directives within the analyzed markup.
Capability inventory: The logic defined in SKILL.md and references/rule.md is restricted to analysis and reporting; it does not include instructions for file system modifications, network requests, or shell command execution.
Sanitization: There are no instructions for sanitizing or filtering the input data before processing.

aria-hidden-focus