The run's report

A run’s page is its report: the verdict and the counts at the top, then every result as a row. Above the rows stand the filters and the switch between List and Suites; under the run’s tags, what the run is for. In the head, beside Delete run, a manual or mixed run can be launched again.

A long run’s list grows as you scroll. The page shows the newest hundred results and reads the next hundred as you come near the end, and the result panel’s ‹ › step on into them. The counts at the top are always the whole run’s. With a filter on, and in Suites, the page reads everything the filter leaves, so «4 of 120 shown» and a group’s number count all of it.

Filter the results

• A chip for each status the run holds, with its number — passed 112, failed 3, broken 1. A status nobody got has no chip. Several chips on are any of them: failed and broken together show both. The chips count what the line of counts at the top counts: a muted case’s failure is not under failed or broken but under a chip of its own, muted 2 (A muted case’s failure).
• With defects — the results that carry a link to a ticket in your tracker — narrows what the status chips leave.
• Search by case name finds any piece of the name, in any case, 300 ms after the last key. Esc clears it at once.
• Sub-status ▾, when the run’s results carry a sub-status: the names under their statuses — what went wrong first, then blocked, skipped and passed — each with how many results carry it. A tick applies at once; several ticked are any of them, and with a status chip on, both narrow — failed and Product bug shows the failed results put down to the product.

Different filters add up: failed with the search «coupon» and With defects shows the failed results whose case’s name has «coupon» and that have a link. The line under the filters says what is shown — «4 of 120 shown · failed or broken · the name has «coupon»» — with Clear filters. The counts at the top stay the whole run’s.

When nothing matches, the list says so — No result matches these filters. — with Clear filters.

A result with a sub-status shows it after its badge, in the row’s quieter tone.

The sub-status after the badge. A result without one shows nothing there.

«Sub-status ▾» groups the names under their statuses, failures first. One ticked, the line under the filters names it.

A name the address asks for that the run no longer carries stays in the menu, ticked, so it can be taken off.

An open manual run

While a manual run is open its rows are the ones you mark, so it has no filters: a narrowed list would hide marks already given. They appear once the run is finished.

The address is the filter

The filters and the view live in the page’s address, so a link opens the same report, and Back and Forward step through what you asked:

/runs/<id>?status=failed,broken&defects=1&q=coupon&view=suites

A sub-status is a name, and a name may hold a comma, so each ticked one is a subStatus of its own: ?subStatus=Product+bug&subStatus=Environment+issue.

While it asks

The filtering happens in Plune, not in the browser, so a chip turns on at once and the list stays where it is, dimmed, until the answer comes — the line under the filters says Loading…. Only the answer to the last question is shown. If the question fails, a card above the list says so with Try again, and the chips and the list stay as they were.

List and Suites

List shows the results as they came. Suites groups them by where their case sits: one group per suite path — e2e/cart.spec.ts › Coupons, or Checkout › Payment for a person’s suites — so where it failed reads at once. A spec’s file stands in for the folders above it. Groups with a failure come first, open, with «2 failed · 1 broken»; the rest follow by path, closed. A group’s head opens and closes it, and the filters work inside the groups.

What the run is for

Under the tags, Add a description — Edit description once there is one — takes plain text up to 2 000 characters: what the run checks, what to watch, what is known to be broken. Save on an emptied field removes it. An owner or a member writes it, on an open run or a finished one; a reader reads it.

Launch it again

Re-run the failed cases

Re-run N failed on a manual or mixed run starts a new manual run of the cases that failed or broke — the whole run’s, whatever the filters show, and not a muted case’s failure. The dialog names them and the new run, «release regression · re-run» by default. Each case goes back to whoever ran it, or to the run’s own assignee for a case nobody marked; no mark comes along, so the new run starts unmarked. Start run opens it.

Launch a copy

Launch a copy on a manual or mixed run starts a new manual run of the same cases — every case the run expected or has a result of — assigned as this one was, «… · copy» by default. No mark comes along. Both keep the run’s environment and labels, and leave out a case that went to the Bin since.

An automated run

Plune does not start CI, so on an automated run Re-run N failed shows the command that runs the failures where your tests run — a terminal or a CI job — with Copy command:

npx playwright test --grep "$(plune run grep <run-id> --failed)"

plune run grep (CLI 0.21.0) prints the run’s failed and broken tests as one --grep pattern, as plune plan grep does for a plan, and the reporter sends their results as a new run. A muted case’s failure stays out, as the platform counts it. Only the run’s id goes into the command, never a test’s name.

Export the run

A run exports as XLSX or PDF. A finished export stays in the AI operations list.

The XLSX has three sheets, in this order:

• Summary — the run’s name; the counts of passed, failed, broken, blocked, skipped and muted; Share passed, written «passed / all (percent)»; the number of results with attachments; and the top 10 failures.
• Results — ten columns: Test case, Suite, Status, Muted, Sub-status, Duration (ms), Executed by, Error, Attachments, Recorded at.
• Attachments — one row per attachment, with a real clickable hyperlink. It contains no formulas.

The PDF carries the same summary on its first page.

The file is kept for 24 hours. After that the row shows Expired for 7 days, with an Export again button that works from any browser.

Who may do what

	Owner, member	Reader
Filters, search, List and Suites	yes	yes
Write the description	yes	reads it
Re-run N failed, Launch a copy	yes	no — the API answers 403

The API

• GET /v1/runs/{id}/results answers a page at a time, newest first. Without limit you get the newest 100 results, plus next while more remain; pass next as before for the page after it. limit asks for another size, from 1 to 1000. Anything else is a 400: { "error": "`limit` must be an integer from 1 to 1000" }. A page can also end early, with next, when its error texts are very long. To read the whole run, follow next until an answer has none. One answer’s length is not the size of the run: GET /v1/runs/{id}/summary counts the run. The filters below ride beside limit and before on every page.
• GET /v1/runs/{id}/results takes status — ?status=failed,broken, or the key repeated — and q, a piece of the case’s name in any case, next to defects=1; different kinds add up. status is the result’s own; with muted=true or muted=false it is read the way the chips read it (The API). Each result carries suitePath, the suites from the root down to its case ([] at the root), and subStatus.
• GET /v1/runs/{id}/results?subStatus=Product%20bug&subStatus=Environment%20issue — the results with any of these sub-statuses, as written; with status both narrow. A name under two statuses matches both. GET /v1/runs/{id}/summary answers subStatuses: each name the run’s results carry, with its status and count, in the order of the statuses — what «Sub-status ▾» offers.
• PATCH /v1/runs/{id} takes description: up to 2 000 characters; null, an empty or a blank text removes it. A finished run takes it when it is the only field in the body.
• POST /v1/runs/{id}/relaunch with { "cases": "failed" } or { "cases": "all" } and an optional title starts the new manual run and answers it with 201. An automated or eval run answers 409, as does a run where nothing failed; a new run counts against the daily run quota (429 when it is spent). The source run’s history records which run it became.