Changelog

We now return a 503 when our service is briefly unavailable, and a 502 when a response can't be parsed. Both replace the old generic 500. Your retry logic can trust the status code now.

Proxy Finder now keeps its ranking data across restarts. After a deploy, your requests route through proven-good proxies right away instead of waiting for the picker to warm up again. The first-request slowdown that used to follow restarts is gone.

Proxy Finder pulls in newly validated proxies incrementally now, so fresh IPs land in rotation seconds after they're verified. Stale ones drop out on the same cadence. If you've seen the occasional retry on a proxy that just died, that window's smaller.

Single, Proxy Finder, and Browser each get their own filter chip on the Overview tab. Pick one to see just that product's summary, timeline, and outcomes. Activity rows now carry matching colored symbols so you can scan the table fast.

When your login cookie expired or two tabs raced the callback, you used to land on a raw white error page. Now we wipe the stale cookies and send you back to a fresh login form. Other auth errors (token failures, account issues, expired support sessions) get a proper branded page with a retry button instead of bare text.

Two fixes for the login flow. Pinch-zoom works again for low-vision users (the viewport no longer caps zoom at 1x). And the theme you picked elsewhere on foura.ai now carries through to login, signup, and recovery pages instead of resetting to your OS preference.

When you create your first API key, the success modal shows a working curl with your real endpoint and X-API-Key header. The Overview page has sample data ready so you're not staring at an empty dashboard, and the API Keys list now shows a proper empty state with next-step hints.

Toggle light or dark on one FourA page and every other open tab switches with it. The theme cookie is the single source of truth now, so the app, blog, docs, and status portal all stay in sync without a refresh.

Poll FourA's live service health without auth: there's a new /api/v1/status on updates.foura.ai that mirrors what the status page shows, cached for 15 seconds. The homepage indicator on foura.ai uses it too, so it's finally showing real state again.

Hover a day on updates.foura.ai and you'll see downtime split by severity level. Uptime percentages now skip days where we had no data instead of counting them as gaps. Public incident titles read cleaner too, without the em-dash clutter.

The status page now shows how much of each day was degraded: hover any day bar for the exact minutes. Dates we weren't yet monitoring a service render as grey instead of green, and a three-state legend under the banner explains each color.

FourA's proxy listing dropped from 460ms to under 2ms for a typical 200-row page. We've trimmed unused fields from the response, rebuilt the sort index, tuned the database for SSD storage, and cached filter options between requests. Proxy picking stays snappy even when admin tools are pulling data at the same time.

Your dashboard's Rate Limit column on the by-domain and top-clients tables had been rendering as "--" and clicking the header to sort did nothing. Both work correctly now. If you've been trying to spot rate-limited traffic across your domains, the numbers are finally there.

If the verify email didn't show up right away, some people clicked resend a few times and got four or five copies. We added a 60-second cooldown on the resend button and a 30-minute server-side cap. One resend is enough now, and your inbox stays clean.

The service health section has a cleaner layout. No more flash of empty content while status data loads, and status indicators align properly.