This skill turns Pageres into a narrow operator workflow: an agent takes the same URL, captures it at multiple viewport sizes, and returns a clean set of screenshots for visual QA. The useful behavior is not “use a screenshot tool” in the abstract. The useful behavior is: given a page or a short list of pages, capture desktop and mobile views consistently so a human can inspect layout drift, hero changes, broken sections, or release regressions without opening a browser and resizing windows by hand.

Invoke this when the job is specifically about visual verification. Good fits include checking landing page changes before publish, building before-and-after artifacts for a design review, generating proof screenshots for client approval, or creating a lightweight visual baseline after a deploy. Do not invoke it when the user needs general site browsing, interaction testing, scraping, or accessibility audits. Pageres is the boundary here: it captures pages, it does not replace Playwright, Selenium, or a full QA suite.

The scope boundary matters because it keeps this entry skill-shaped instead of turning it into a generic product listing. An agent using this skill should accept URLs, viewport presets, and output naming rules, then produce organized screenshot artifacts. It should not pretend to do end-to-end browser automation. Integration points are straightforward: URL lists from a CMS or deployment pipeline, file output to local folders or cloud storage, and handoff into review docs, release notes, or issue threads.