Changelog

• Test runs lists every run the agent starts. A run is filed as soon as it ends. Before, a second run in the same chat turn, or a run that finished after its turn, never appeared. Stopped runs, and commands that aren't test runs, aren't filed.
• Terminals: keys typed quickly reach the shell in order. Hiding the bottom panel keeps every terminal and whatever is running in it, and no shell starts until you first open the panel.
• Live Browser: popups and new tabs identify as Chrome rather than HeadlessChrome from their first request, and a page's native dropdown lists show over the live view.
• Git: the header Git popover closes on Escape, and Source control updates as soon as a sync from the header button finishes.

Released as testrelic-cli 1.7.2.

Studio and the CLI now run the Cursor engine whenever Cursor's CLI is installed and signed in.

• Windows installs are found, and start. A native install from irm 'https://cursor.com/install?win32=true' | iex is found even when your PATH predates it, and its first message no longer fails to start. The install hint now gives that command, and Studio's engine picker lists Cursor as available.
• Plan mode works in a new folder. A plan turn in a folder Cursor hasn't trusted used to stop with "Workspace Trust Required". It now runs, on Cursor builds that support --trust.
• WSL works too, with Cursor installed in ~/.local/bin by its install script.
• Token usage is counted, cache reads and writes included.
• New names and overrides. Cursor's newer agent command name is recognised, and TESTRELIC_CURSOR_BIN points TestRelic at a specific Cursor binary.

Released as testrelic-cli 1.5.3.

Artifacts render as designed in the installed Studio, and a test's Replay now agrees with its Overview.

• Artifacts look as designed. In the installed Studio, HTML artifacts used to render with browser defaults: plain serif text, unruled tables and collapsed stat tiles. They now keep their own styles, along with Studio's document defaults for titles, headings and tables.
• Replay lists the Overview's steps. A test's Replay tab now lists exactly the steps its Overview shows, with the same failed and didn't-run marks. That includes tests with no trace, and runs the agent started, which used to show different steps in Replay or none.
• Steps line up with the video. Clicking a step seeks to the right moment in the recording, including for agent runs.
• A notice when a run can't be replayed. If a run recorded no steps, or didn't record when they happened, Replay says so and offers Re-run this test. You can dismiss the notice for that run.

Released as testrelic-cli 1.5.3.

The agent chat can now answer with interactive cards as well as text, and you choose who it answers for.

• Cards in the conversation.
  • The agent's answers can render as cards: test cases, bug reports, checklists, run results, test flows, coverage, test plans, locators, API calls, metrics, charts and next steps.
  • Test cases export as CSV that TestRail, Xray, Zephyr and qTest import, or as Markdown for a ticket.
  • A card shows only what the agent actually gave, and one still being written shows a placeholder. Cards work under every engine.
• Manual or Automation. A switch in the chat header sets what answers lead with. Manual puts test cases, checklists and bug reports first, and Automation leads with runnable code, locators, runs and API calls. The empty chat's starter cards follow your choice.
• Clearer turns. Code comes in highlighted cards with the file named at the top, plans show their progress, and file edits show lines added and removed.
• Artifacts. An artifact opens with a Preview / Source switch, and a deck opens in a slide viewer. An artifact's card in the chat has Open, which takes you to that artifact.
• Trace opens Playwright's Trace Viewer. A test's Trace tab now shows Playwright's own Trace Viewer for the run's trace. If the viewer can't start, the tab says so and offers Try again.
• Fixes.
  • Every picker in the composer's Chat settings now opens properly instead of collapsing: assistant, model, reasoning effort and default run target.
  • Esc closes the open menu instead of stopping the reply.
  • The terminal's colours follow light and dark mode and stay readable in both.

Released as testrelic-cli 1.5.0.

Studio's Cloud mode is rebuilt on the Cloud Platform's own layout, look and words, and a change made in one now shows up in the other.

• The same navigation as the web. The sidebar holds Ask AI, Repositories, Test Runs, Monitoring, Alerts and Artifacts. A repository opens with the web's header and tabs: Overview, Test Cases, Test Runs, Test Plans, Monitoring, Memory, Integrations, Alerts and Settings. Lanes and Telemetry are now under the repository's Settings.
• Two-way sync. Acknowledge an alert in Studio and the web refreshes and says so, for example "Alert acknowledged in Studio by …". Edit a lane on the web and Studio redraws. Neither needs a reload.
• From the cloud to your code. A cloud test offers three actions:
  • Open in Code;
  • Run locally, which runs the matching test in the open folder, and says so instead of guessing when it can't find one;
  • Debug in Live Browser.
• No dead ends.
  • Cloud mode opens on your last repository, and the time window starts at Last 30 days.
  • A screen that fails to load says so and keeps what it last showed.
  • A change that fails offers Retry.
• The platform's real data. Alert names, modules, trends and ingest status show what the platform holds. Runs and repositories page past their first page, and failure text no longer shows terminal colour codes. Clicking a step in a test's session seeks the video to the right moment.

Released as testrelic-cli 1.5.0.

Tests in Studio is now a table you can select from, copy, group and export, like a spreadsheet.

• Select and copy cells. Ways to select:

  • click a cell;
  • Shift+click or Shift+arrow keys for a block;
  • a row number for the whole row;
  • Ctrl+A (Cmd+A on a Mac) for everything.

  Ctrl+C copies the selection as cells that paste straight into Sheets or Excel, and Ctrl+Shift+C adds the column names. While cells are selected, the footer totals them: tests, passed, failed, total and average duration.

• Columns you control. The columns are Test, Status, File, Folder, Type, Tags, Project, Duration, Retries and Error. Sort from a header, and resize a column by dragging its edge. The Columns menu shows and hides columns and turns on compact rows. Row numbers and the Test column stay in place while the rest scrolls sideways.

• Group up to three levels deep by Folder, File, Describe, Type, Status, Tag, Project or Modules. Add a level from the group bar, or drag a column header into it. Each group row shows subtotals, and a totals row closes the list. Until you change it, the list groups by Folder, then File.

• Export CSV saves the visible columns, in their current order, to your Downloads folder.

A single click now selects a cell instead of opening the test. To open a test, double-click it, press Enter, or use Open when you hover the row.

Released as testrelic-cli 1.5.0.

Open a test in Studio's Tests view and it now says what happened in plain words first. One tab strip then takes you into the evidence.

• A result card. The Overview opens on a sentence about the outcome and the step the test broke at. Below it are a strip of its last 10 runs, labelled New failure, Keeps failing, Fixed or Stable, or with how often it needed a retry, and the duration compared with its usual time.
• One "What happened" list. It replaces the Workflow, What went wrong and Compare tabs. Steps after the failure read "didn't run". The failed step opens in place, with Expected, What it got and Waited on, the code, Watch this moment and Copy bug report.
• Flaky and expected failures read right. A flaky test says which try passed and shows both tries side by side. A test.fail() that failed as intended reads as a pass.
• Tabs flag what needs a look, such as "1 error" on Console or "1 failed" on Network.
• Replay puts a tick on the scrubber for each step. It adds previous and next step, speed, fullscreen, a jump to Where it broke, and an All steps / Assertions filter.
• Console groups lines by the step that was running. It has level filters, Copy all and Ask AI about this error.
• Network adds type filters, search, sorting and a details panel.
  • Share this request copies a request as cURL (bash / zsh or PowerShell) or as an AI prompt. Hide cookies and tokens is on by default.
  • Ask AI in Studio puts the prompt in the chat box without sending it.
• Files previews each file, with Show in folder and Copy path. Eval cases shows every case against every metric in one grid, each score drawn against its threshold.

Released as testrelic-cli 1.4.0.

The Live Browser is steadier when memory runs short, after a crash, and when an agent hands it a half-written address.

• A low-memory warning in the chat header. When the computer is short on memory, the agent chat's header shows Memory low (under 1 GB free) or Not enough memory (under 512 MB). Hover it for what to do, or click it to check again. A browser that fails to start says so when low memory is the likely cause.
• Addresses load the way you meant them. An address with only one sensible reading is completed: localhost:3000 opens http://localhost:3000, example.com gets https://, and 0.0.0.0, where a dev server listens, opens 127.0.0.1. Anything ambiguous, such as : or /checkout, is refused with a sentence naming an address that would work. The same rule covers the address bar, the agent's navigation on every engine, and your project's app URL.
• Agents stay connected through a recovery. If the browser crashes and comes back, an open chat keeps driving it.
• Idle browsers close. A workspace's browser closes 5 minutes after its workspace goes off screen, provided no pane is showing it, no agent is connected and no other window shares it. When memory is low, it closes after 30 seconds.
• Stop stops the tools too. Stopping a turn on the TestRelic Cloud or Ollama engine also cancels the browser and MCP calls it still had running.
• Typing and touch reach the page as they should. Keys arrive as real key presses, including F-keys and AltGr, and with a mobile device preset on, presses and drags are sent as touches.
• Fixes. The DevTools panel, which could open blank, builds its panels again. A fresh launch no longer reopens the last session's tabs. On Linux, the pane no longer freezes after its first frame in a desktop session. On Linux and macOS, the browser keeps its cookie key in Studio's own profile, so it never asks to unlock the desktop keyring or Keychain. Existing profiles move over once and stay signed in.

Released as testrelic-cli 1.3.0.