New audit

Point it at a URL, choose what to check, start. The report updates live while it runs.

Target

URL Discovery starts here: seed page, sitemap, then a shallow crawl of same-site links.

Crawl & viewports

Viewports Comma-separated WIDTHxHEIGHT. Screenshots and responsive checks run once per viewport.

Max pages Crawl cap. Empty = default (20).

Max depth Link depth from the seed URL. Empty = default (2).

Advanced: mask selectors

CSS selectors to mask in screenshots (one per line) Hides dynamic areas (carousels, chats, ads) so screenshots and Figma diffs stay stable.

Delivery checklist

Profile

Adds the pass/fail delivery checklist to the report.

Expected production domain For the “right domain” delivery check.

Site should be indexable — apply the robots.txt / sitemap checklist items.

Access & login

Extra request headers (for gated staging, e.g. X-Staging-Key)

No login

HTTP Basic

OTP / magic link

Figma comparison

Find the file by name

Looks the file up through the Figma API and fills the key below — no need to open Figma.

Figma file Enables screenshot-vs-design comparison per page.

Auto-match frames by naming convention

Page → frame mapping (optional — for frames auto-match can't find by name)

Advanced: mapping as JSON

Mapping (JSON list) Same shape as manage.py figma_mapping_template produces; merged after the rows above.

AI review

AI review layer — screenshots + findings reviewed by a vision model; enriches the fix guide.

Only review pages that already have findings (cheaper).

Automation

Callback URL Gets a compact JSON summary POSTed when the run finishes.

Advanced: E2E flows

Flows (JSON list) Run by the browser-checks module on each page: goto / click / fill / expect steps.

Runs on the worker queue — you land on the live report immediately.