Ariada Browser Extension — Documentation

What this module is

The Ariada browser extension is a product surface, not an engine: it is a thin Chrome side-panel UI that calls the same open-source scan engine (@ariada-org/core-engine via the in-page adapter core-browser) the CLI and CI Action use. You open a page, click Scan this page, and it reports findings across six compliance domains in a grid — no account, no upload, no server.

Is this a design mockup? No. It is a real, buildable extension in packages/ariada-extension. Every screenshot below is the actual panel rendered by the end-to-end Playwright harness, not a Figma comp.

Type: Browser extension (Chrome side panel + on-page launch button)
Where it lives: Source in the monorepo (packages/ariada-extension); ships to the Chrome Web Store as the consumer install target
What it runs: The six OSS domain analyzers on the current document, in the browser; cross-origin pages can't be fetched from a page context, so additional URLs are handed to the companion CLI
Status: Built & test-covered surface; Chrome Web Store listing not yet published (gap)

Who uses it & why

Role	How they use it	Value
Front-end / QA developer	Quick in-browser check of the page they're building before opening a PR	High — zero-setup, no CLI/CI needed
Accessibility engineer	Spot-audit any live page across 6 domains, export JSON	High
Compliance / procurement reviewer	Sanity-check a vendor or public-sector site against EAA-scoped rules	Medium-High
Site owner (non-developer)	One-click "how bad is my page" with plain findings	Medium

Strategically the extension is a top-of-funnel adoption driver: the lowest-friction way to experience the scanner, designed to convert curious users into CLI/CI adopters and, eventually, paying .ai customers.

Distribution — free and open source

The Ariada browser extension and the scan engine it runs are free and open source under the EUPL-1.2 licence. There is no account and no upload to use the in-panel single-page scan across all six compliance domains.

Capability	Availability
Single-page six-domain scan in the panel	Free · open source
Local JSON export of findings	Free · open source
The scan engine, rule packs and EAA evidence emitters (npm)	Free · open source
Companion command-line tool for batch / cross-origin scans	Free · open source

Everything documented on this page is part of the open toolkit. The project is funded as public-interest open source.

Getting started

Discover — Chrome Web Store listing and npm.
First value — install, scan the page you're on, see six domains of findings in seconds.
Habit — use it in daily development and QA; export JSON; add the command-line tool for batch scans.
Integrate — wire the scan into continuous integration with the companion Action.

Where it will be: the Chrome Web Store (primary), with the same open core reused by the editor extension and the in-browser surface. The store listing and assets are a pre-launch gap.

Documentation & help — current state

Docs site: An Astro Starlight site exists at apps/docs (framework decided in an ADR); per-package usage pages are mostly unfilled — the extension needs a quick-start + permissions explainer page.
Package README: Present for the extension; most sibling packages still lack one (npm landing pages are bare) — flagged as the top docs follow-up.
Recommended form: One Starlight "Browser extension" guide (install → scan → read findings → export → when to use the CLI), plus a 60-second store demo clip and a permissions/privacy note (local-only, no upload).

AI chat & digital-adoption helpers — plan

Honest status: there is currently no in-product AI-chat or digital-adoption-platform (DAP) integration, and no committed plan in the repo. The following is a proposal, not existing work.

In-panel AI assistant — "explain this finding", "show me the fix", "draft the accessibility statement" — calling a managed model only on user action (kept off the deterministic scan path, consistent with the no-AI-in-core design).
Digital-adoption walkthrough — a first-run guided tour and contextual tooltips (a lightweight DAP layer) so a non-developer understands each domain column.

Screenshots — click any image to open it full size

From the real Playwright end-to-end harness at 1280 px (the deployed side panel is ~400 px wide). Each thumbnail opens a full-size, zoomable overlay.

Side panel idle: ariada scanner header, blue Scan this page button, six domain checkboxes checked, Download JSON disabled — 1 · Idle — ready to scan, six domains selected

Report grid after scanning one site: seven columns — Site, Accessibility, Privacy, Security, AI readiness, Structured data, Sustainability — 2 · Result grid — one site, six domains

Report grid with two sites and cross-domain interactions listed below — 3 · Two sites + cross-domain interactions

Error state: red message — cannot scan chrome:// internal pages, only http/https — 4 · Error state — unscannable page

Settings page: six active domain modules each with a checkbox, add-a-module input — 5 · Settings — domain modules

Settings with a valid companion-CLI module name and green validation feedback — 6 · Settings — module validation

Assessment summary

Full UI assessment (2026 best-practice audit, competitor screen-to-screen, user pains, recommendations) is in assessment.html. Headline: the panel passes its own axe-core gate (the scanner must be accessible — "cobbler's shoes"), with open gaps around a 375 px mobile/side-panel-width capture and the unpublished Web Store listing. Competitors compared: axe DevTools, WAVE, Lighthouse, Accessibility Insights.

What we test: idle / scanning / result / error / settings states; keyboard + focus; empty & error states; and an axe-core self-audit of the panel itself.