The Agent Skills Directory

[SAFE]: The skill's logic is implemented entirely with native browser APIs such as TreeWalker, requestAnimationFrame, and CSS custom properties. No external scripts are loaded, and no network requests are performed. All documentation links in REFERENCES.md target well-known and reputable services.- [INDIRECT_PROMPT_INJECTION]: The skill processes text nodes from the DOM to apply visual reveal effects, creating a minor surface for data ingestion. 1. Ingestion points: Text content is read from elements marked with data-reveal via document.createTreeWalker in demo/index.html. 2. Boundary markers: The documentation instructs developers to maintain an untouched accessible text source in the DOM for screen readers. 3. Capability inventory: The skill is limited to visual styling and DOM restructuring (wrapping words in spans). 4. Sanitization: The implementation uses textContent to populate generated word spans, which prevents cross-site scripting (XSS) or other injection attacks from the source text.

scroll-scrubbed-word-reveal