Run a whole project

Run every saved request or one test group and read the summary.

In short

"Run all" runs every saved request of a project, or one test group, and sums up what passed, what failed and what was slowest. The board then shows your coverage: what was tested, what failed and what never ran.

Before you start

Who: "Run all" needs Run on the project. A request that would run against Production at the Run level is listed as not run.
Plan: every request counts toward your allowance (see /help/settings/plan-limits). A request that is switched off is left out and not counted; tick it on the board to run it anyway.
Editing while it runs is safe: the run writes only each request's result, never over what you saved meanwhile.
First: check the inherited authentication and the environment's base URL.

Steps

Open the project board and check that inherited authentication is right.
Choose "Run all", or run one test group.
Wait for the summary: passed, failed, slowest, and the names of failed requests.

Result: every result is stored, and the board shows how many requests are tested, failing, slow and never run.

Read the board

The counts sit in one line under the project title, such as "12/20 tested · slowest 840 ms". The filter chips on the strip show from 20 requests, or while one is picked.

A long project opens fast: each test group shows its first 50 requests, with "Show more" under it for the next ones, and its count is how many it holds under the search and the chip. The search, the chips and the column you sort by work on the whole project, not only on the requests on screen. In a folder or a team list, "Run all" starts one run per project and follows them all until the last one ends.

Good to know

Requests go out four at a time and each gets 30 seconds before it is recorded as a failure. One broken URL does not stop the rest.
A request passes when the call completes, the server answers with a status from 200 to 399, and none of its checks fail. Redirects are not followed, so a 3xx counts as a pass.
Per request: a default environment, the OAuth inheritance, "Enabled in test suites" to skip it without deleting it, and its own response-time bands.
"Run in CI…", behind the chevron beside "Run all", shows the command and pipeline steps that run the project with the {{brand}} CLI. See /help/api-testing/run-in-a-pipeline.
Below 768 px the requests table lays itself out as labelled cards, with the request's name as the card's title. Run and the "…" menu stay on every row.

Run all can fail what Send passes Four calls at once may hit a rate limit, a lock or a full connection pool. That is a real finding, not a false alarm.

If something goes wrong

Many requests fail at once: check the inherited authentication and the environment's base URL first.
"Environment '...' needs a base URL to run relative URL": set the environment's base URL.
Requests are listed as not run: your level is Run and the environment is Production.

/help/api-testing/time-bands
/help/api-testing/test-suites
/help/api-testing/run-in-a-pipeline