Changelog

Ask AI conversations running in Studio or the CLI now show up in Ask AI on the web, where you can follow them as they happen.

• A Chats / Remote switcher heads the Ask AI sidebar. Chats holds your web conversations, as before. Remote lists conversations from Studio and the CLI: live ones under Live now, ended ones under Archived. Search works inside whichever list you're on.
• Watch a turn as it streams, including the artifacts it produces along the way, rather than after it finishes.
• Each session shows where it came from, Studio or the CLI, and what state it's in.

Nothing to set up: conversations from Studio and the CLI appear on their own.

Also in this release, signed-in browser sessions are protected against cross-site request forgery. API keys, the CLI, the SDKs and webhooks work exactly as before.

A test run in Studio can now be stopped, and the Tests section has been redesigned around one primary action per screen.

• Stop a run. While a run is going, the Tests list shows Stop beside "Running…", and a test's detail view offers Stop run in place of Run this test. Stopping ends every process the run started — the runner, its workers, the browsers and the dev server Playwright launched — and the list goes back to the results it had. It reads as "Run stopped", not as a failure, and records nothing.
• Closing Studio mid-run stops the run too, the same way. A run the assistant stopped reads as Stopped in the chat, and never picks up an earlier run's report.
• The CLI cleans up after itself. When the assistant's background test runs are stopped, or the CLI exits or is interrupted, every process those runs started is shut down: politely first, then for good.
• One look across Tests. The list, a test's detail, Needs rework and Test runs share one design. The list scrolls in any window, and every Tests view fits a narrow one.
• The list matches what Playwright runs. It follows your config's testDir and testMatch, finds every spec file (and the Tests search finds a file by name), includes tests written against a custom fixture, and no longer lists a helper call inside a test as a test of its own.
• Results you can trust. A test.fail() that fails as expected counts as passed. Every test keeps its own evidence across reruns, and says why when that evidence can't load. A comment inside your reporter array no longer makes Studio add its reporter twice.

Released as testrelic-cli 0.109.0.

Two tests sharing a title in one file used to collapse into a single row in Studio's Tests list: selecting either one selected the first, opening the second showed the first one's steps, and rerunning by name ran both. Studio now identifies a test the way your framework does, by file › describe blocks › title.

• Each test is its own row, with its own declaration line and its own steps.
• A rerun targets file:line, so it runs the test you picked and nothing else.
• The same title under two different describe blocks is no longer mistaken for a collision.
• A real collision is explained. A band says what the shared name costs you in your framework's own terms, and a Fix the shared name rail hands the rename to the agent.

Released as testrelic-cli 0.106.0.

Studio's agent chat could get stuck reporting "your sign-in has expired" even right after you signed in — and pressing Sign in again didn't clear it. It does now.

• A lapsed session heals itself. When a chat turn is refused for authentication, it re-checks your saved sign-in and retries — so a sign-in you completed while Studio was open is picked up on your next message, with no restart.
• Sign in again actually works. Signing back in through Studio now refreshes the credential the chat uses, instead of leaving the conversation stuck.
• The message is honest. When a sign-in genuinely has expired, the prompt now points only at steps that resolve it.

Released as testrelic-cli 0.105.2.

If your CLI or Studio is running a version too old to update itself, TestRelic now says so the next time you use Ask AI — on the one channel a stuck install can still reach — with the single command that fixes it.

• The notice reaches both the terminal CLI and Studio's agent chat, since Studio drives the same CLI underneath.
• It's shown once, not on every turn, and it names the exact reinstall command for your platform.
• It moves with each release on its own. A client merely a version behind — one whose updater still works — gets a quieter "a new version is available" note instead of the reinstall prompt.

No action needed if you're current; you'll never see it. Shipped on the Cloud Platform.

Auto-update is reliable again — and when it can't run, it now says so instead of failing in the background. Older installs that had quietly stopped updating are the reason this shipped.

• Updates survive a download source being unavailable. The CLI and Studio now try more than one place to fetch a release, so a single origin going down no longer strands your install on an old version.
• A new testrelic update command. Run it any time to update on the spot; testrelic update --check reports your version against the latest and which download source answered, and changes nothing.
• A failed update is visible. When an update can't complete it now tells you why and what to do, rather than starting the old version as if nothing happened.
• Bigger releases download reliably. The updater no longer gives up early on a slow connection, and shows real progress while it works.
• Alongside: a round of Studio and CLI interface fixes — sign-in screens you can move and close, a live countdown and a clear way out while a sign-in is pending, and failure states that say what went wrong instead of showing an empty "nothing here yet" screen.

Released as testrelic-cli 0.105.0. If your install is older than 0.104.0 it can't reach this fix on its own — reinstall once from the install steps and it will keep itself current from then on.

The reporter learned to say which element caused a visual failure two days ago. This release closes the loop: the CLI and Studio show the same attribution, and a reviewer can accept or reject the new baseline without leaving the terminal — with a record of who decided what, and why.

• tr visual accept / tr visual reject. Accept copies this run's render over the committed baseline file the report names — the one Playwright reads next time — and nothing else; reject changes no files. Both refuse what cannot be decided (a pass, a baseline that was only written, a size mismatch), both record the decision as a line in <report>/visual-decisions.jsonl before anything touches the network, and both sync it to TestRelic Cloud when the run was uploaded there — otherwise one line says why it stayed local. --note travels with the decision. See Visual regression.
• tr visual lists a retried test once, with the final attempt's verdict, and names a comparison by snapshot, by <spec>::<snapshot>, or by test id — an ambiguous name is refused with the candidates listed, never guessed.
• Studio has a Visual tab in the cloud test stack, beside Steps, Console and Network: baseline against this run, side by side or as Playwright's pixel diff, the element list with the changed element's box drawn over the render, and Accept / Reject wired to the platform. The same tab appears for a report opened from disk, without the decision buttons — nothing leaves the machine there.
• The step cursor that scopes the panels is set from the Steps tab in the cloud test stack.

Released as tr 0.103.0 and the matching Studio build. Accept needs a report from @testrelic/playwright-analytics 2.16.2 or later, which records where each committed baseline lives.

When something goes wrong, Studio now tells you what happened and how to fix it.

• Every error comes with steps. A short explanation, then what to do, in order. The technical detail goes to a log file instead of filling the dialog.
• Every error has a page. Click More help in the message to jump straight to the matching section of the new Troubleshooting Studio page.
• Two common failures are named properly now. Antivirus software that inspects HTTPS traffic can break Studio's connection — Studio recognises that and retries on its own. And a mistyped server address used to be blamed on antivirus, which no antivirus setting could fix.