PixyScan changelog

PixyScan - 24 August 2026

Mon, 24 Aug 2026 09:00:00 GMT

Retention became a plan feature, alerts got a home per site, and the contact form started reaching a person.

Added: Configurable data retention per workspace plan - 30 days on Free, 180 on Hobby, a year on Basic, two on Pro, and up to ten by contract.
Added: A site-specific alerts section, so scan and score notifications are readable per property rather than only as one workspace-wide stream.
Added: Favicon probing and caching, so a site list shows the sites rather than a column of identical placeholders.
Changed: The AI readiness panel now prints the weight beside each of its three factors, so a low score names the group that produced it instead of showing a gap with no cause.
Changed: Invitation handling and per-site permissions tightened, so a role genuinely bounds what a member can reach.
Changed: Clearer messaging on the billing screen about what is and is not yet chargeable.

Fri, 21 Aug 2026 09:00:00 GMT

The billing screens were rebuilt, and link health learned about inbound links.

Added: Inbound link tracking on the link health view - which pages link to a URL, not only which URLs it links out to.
Added: Billing statistics with usage clusters and period pace, so you can see whether this month's credit will last the month.
Changed: The Plan and Billing pages redesigned around stat tiles, with the retention window shown as a plan limit.
Changed: The affected-pages table on an issue's detail screen rebuilt - the list of URLs a finding was recorded on is the thing people open an issue for.
Changed: The site rail shows the health score as a dial rather than a figure, which reads faster and is legible at a glance.
Fixed: Role-based permissions on the scan action, with tests - a viewer could reach a control they could not use.

Thu, 20 Aug 2026 09:00:00 GMT

Crawl control got finer, for sites where one set of rules does not fit every check.

Added: Per-check URL scoping: whitelist or blacklist the pages a single check runs against, for the cases where a rule genuinely does not apply to one section.
Added: Rate-limit healing mode in site settings, so a crawl that starts getting 429s backs off instead of failing the run.
Fixed: Crawl depth was reported shallower than it was on some sites.
Fixed: The status-code filter now distinguishes URLs that were never fetched from URLs that answered something.

Tue, 18 Aug 2026 09:00:00 GMT

The score stopped being a number you had to take on trust, and two checks that measured editorial taste were retired.

Added: A health score explainer in the product, breaking the number down by discipline and showing what each unfixed check is costing.
Added: An unsaved-changes guard, so navigating away from a half-edited settings screen asks first.
Removed: The question_format_headings check. How a section is titled is an editorial choice, and scoring it was enforcing a house style nobody asked us to hold.
Removed: The ordered_lists_howto check, for the same reason.

Mon, 17 Aug 2026 09:00:00 GMT

llms.txt started being graded on its contents rather than its existence.

Added: llms.txt contents are audited - whether the file follows the format, and whether the URLs it lists still resolve. Publishing a file with one junk line used to score the same as publishing a good one.
Added: Site-level evidence on findings that belong to the site rather than to a page - robots.txt, the sitemap, a domain-level redirect.
Changed: Scan names in history cleaned up, so a list of runs reads as a list of runs.
Changed: Sitemap entry handling improved, including how a URL's discovery source is tracked.

Sun, 16 Aug 2026 09:00:00 GMT

The audit workspace redesign.

Changed: The audit workspace hierarchy reworked - the route from a score, to a finding, to the URLs it was found on, to that page's full report.
Changed: Filter state is now remembered as you move between screens, rather than resetting on every navigation.