Changelog

Uploaded visual comparisons had nowhere to be reviewed. Now they have two.

• A Visual tab in the session workspace of every Playwright session: every comparison the run made, most changed first, with baseline and this run side by side, a wipe, an onion skin and Playwright's pixel diff. The Elements panel names the element that caused the change and draws its box over the render; the ten things that only shifted below it are filed as knock-on movement, not ten findings.
• A standalone Visual Regression workspace reached the way Web and Mobile are reached, for triaging a whole session's comparisons at once.
• Accept and reject, recorded. A decision is stored against the session with who made it and why; tr visual accept from the CLI lands in the same ledger, so every surface shows the same answer. A pass, a baseline that was only written, and a size mismatch cannot be decided — the workspace says so instead of offering a button that would be refused.
• One row per comparison. Retries and repeat uploads no longer add rows of their own. Each comparison is one row, and the last attempt is the verdict. A first run shows six snapshots as six rows, each marked baseline written, not twelve with three of them green.

• A chat that reached its session token cap used to stop there permanently. It now compacts and carries on — the cap triggers compaction rather than ending the conversation.
• Token spend that was quietly under-reported is now counted in full, so usage reflects what was actually consumed.
• An organization can be given its own Ask AI session token cap and spend ceiling instead of sharing one platform-wide default. The two move as a pair, so a ceiling raised alone can't silently change which limit you hit.

Sending the same file twice for a run — a Playwright retry re-sending a screenshot, or a repeated upload — counted it as a second artifact. The run's evidence then rendered twice, and because usage totals the size of stored artifacts, the same bytes were counted twice.

Re-sending a file the run already holds now reuses the artifact it already has. This affected the CLI, the JavaScript reporter and the Python reporters alike.

When a turn failed, the whole reply was discarded: your question sat in the transcript with no answer, no error and — after a reload — no sign that anything had been attempted, even when minutes of real tool work had gone into it.

A failed turn now keeps whatever the model had already produced, because a partial answer beats a blank, and records why it ended. Turns that finish against a ceiling rather than a failure — idle, wall-clock, token cap or a stuck loop — now record which one, instead of mentioning it only in the prose of the answer.

• A link to a page that doesn't exist now renders Page not found inside the app. Signed-in users were previously bounced to Ask AI, which turned every dead link in the product into what looked like a deliberate visit to Ask AI. Start-page preferences also drop three options that were never real destinations.
• Share, wall, invite and account-setup links carry a token in the URL that grants access on its own. Those tokens are now scrubbed from product analytics and from server logs, and session replay is off on those pages. Run identifiers and route names are deliberately kept, so run analytics and path queries keep working.

• Fixed: selecting a test with no video of its own could show another test's recording instead of an empty state. Fixed in both the Session Workspace and the Ask AI embedded workspace.
• Fixed: the recording's timeline axis was scaled to the run's total duration instead of the selected test's, so a 30-second test inside a 4-minute run rendered crushed into roughly the first 12% of the scrubber.

See Session Workspace for how test evidence is organized.

Where your tests run is now a workspace setting instead of a file on one person's laptop. The new Infrastructure console — now in the header nav, right after Test Runs — manages three kinds of execution target: your own AWS / Azure / GCP machines, a device grid (BrowserStack, TestMu AI), or your own CI pipeline (GitHub Actions, GitLab CI, including self-managed GitLab). A CI target is treated honestly: connecting one provisions nothing, and TestRelic never estimates your CI cost.

See Infrastructure for the full breakdown. Bring-your-own execution targets are a Pro-and-above entitlement, rolling out gradually.

Load testing joins TestRelic as a first-class run type. Upload results from k6 or JMeter and get one workspace regardless of which tool produced the run: synchronized concurrency, throughput, latency, and error-rate lanes, a failure-first playhead, and repo-level service levels that gate or warn on a threshold breach.

See Performance Testing for setup and the full workspace walkthrough.