Live test monitor #

URL: /live/{run_id}

The live monitor renders a streaming view of one in-progress test run. It's the destination for the Launch Test form's Start Test button and the ACTIVE TESTS sidebar block — when an operator clicks either, the browser lands here.

The page is single-run: it doesn't list the run history. For that, go to the Results list. Once the run completes, a "View Full Results →" link appears at the bottom that takes you to /results/{run_id}.

Status card #

A pinned card at the top of the page with two rows:

Status — current state pill. Starts at Starting..., then shifts to an uppercase status (RUNNING, PASSED, FAILED, ERROR) once the runner emits progress / complete events. Colors track state (blue while active, green on pass, red on fail / error).
Run ID — the full run UUID, monospaced.

Below the rows: a horizontal progress bar and the current step label. Both update as the runner emits progress events.

Tab strip #

Tab	What it shows
Events	Streaming timeline of every event the run has emitted so far, in chronological order. Subscribes to the EventStore push channel — new rows append as they arrive.
Channels	Current channel readouts (per-channel last values + sparkline-style recent history).
Output	Tail of the runner's captured stdout / stderr, last 100 lines, monospace on a dark background.

The default tab is Events.

Operator dialogs #

When a test hits a litmus_prompts marker, the prompt renders as a modal on this page. The same session is also pushed to the ACTIVE TESTS sidebar block so an operator on any other page sees an amber row reading "N dialog(s) waiting" and can click straight back here. The modal is the dialog UI; the sidebar is the notifier.

For prompt-design guidance, see Design operator prompts.

When the run finishes #

The Status pill turns green or red based on the runner's exit code.
The progress bar fills to 100%.
The Events tab keeps the full timeline of the completed run.
A "View Full Results →" link appears at the bottom — click to jump to the Results detail page for this run.

The page stops streaming when the run ends; it doesn't poll for new runs.

On stream errors #

If the underlying runner connection fails (OS error, runtime error, value error during stream consumption), the Status pill turns red and reads ERROR. A toast at the bottom of the page prints the exception type and message. The page does not auto-retry — refresh to reconnect.

Underlying data #

The streaming subscription is to the EventStore for the run's data dir, scoped by run_id.
Channel readouts share the same EventStore subscription as the Events tab — pushed updates instead of pull queries.
The captured stdout / stderr buffer is held in-process by the active runner; closing the tab and re-opening this URL re-attaches to the existing stream.

Common tasks #

Start a test and watch it run — Launch Test → Start Test → you land here.
Respond to an operator dialog — when a test pauses, the amber sidebar row brings you back to this URL; the modal opens automatically.
Drill into the completed run — once the Status pill turns green or red, click "View Full Results →" for the per-step detail view at /results/{run_id}.