CAU Agentic Browsing Audit

About

CAU Agentic Browsing Audit helps developers test the parts of a site that a lighthouse agenting browsing audit scan misses. Think dialogs, bottom sheets, popovers, menus, drawers, form errors, hover states, and other UI that only shows up after someone interacts with the page.

Open the DevTools panel, start a session, and capture each state as you reveal it. The extension runs checks in the inspected tab, saves a screenshot of the state, captures the affected elements it can find, and groups the results into a CanAgentUse report.

Use it when you are building or reviewing flows where agents need to click, open, choose, dismiss, and complete tasks. It is useful for modals, menus, checkout steps, onboarding, account settings, and JavaScript-heavy screens where the real problems are hidden behind interaction.

Reports include the captured states, issue severity, affected selectors, element screenshots, fix guidance, and a PDF export for sharing with the team.

The goal is simple: catch interaction bugs before they make your site harder for agents, assistive technology, or keyboard users to operate.