Skip to content

Reading the report ​

bash
npx monkeytest report
npx monkeytest report --json

report prints the latest file in .monkeytest/runs/. Pass a run id to print an older one.

bash
npx monkeytest open

open prints that run's path and the screenshot and WebM paths stored on it. It does not upload them, and it does not fetch the site. There is no hosted report URL. After you change the site, npx monkeytest rerun replays the stored trace and npx monkeytest diff prev latest compares fingerprints.

The outcome line ​

outcome is one of bugs_found, clean, inconclusive, could_not_test.

could_not_test means nothing was tested (no pages, no flows, or the site could not be opened). inconclusive means the site was reached and a real check did not finish. A Playwright error from the runner is inconclusive, not a bug in the site.

A flow is passed only when its outcome check verifies. These two reasons are never paired with passed:

outcome_reasonMeaning
no_assertionThe journey had no outcome to verify. Steps running is not a pass.
assertion_unverifiableA check was present and could not be decided. A missing invented success sentence is this, not a failed flow.
budget_exhaustedThe run finished its time budget before this check. Not a failure to confirm the page.
contact_submit_blockedA contact, enquiry, feedback, support, quote, or get-in-touch message was not sent. Not a pass and not a bug.
request_blockedA sensitive submit or other blocked request left the browser on an error page. Not a pass and not a bug.

When one confirmed bug is filed, outcome_detail is:

text
Found 1 confirmed bug. Open the top one and confirm it on the site before treating this run as healthy.

That sentence is the engine's text. A clean run says the completed journeys had no confirmed bugs. When the cause is the runner, the sentence says "Our test couldn't complete".

login_required is set when a login wall was seen. It is also set on clean and bugs_found when public_screens is 0: every crawled page was auth-gated, and deeper journeys need an account.

Other fields worth reading first ​

FieldWhat it tells you
public_screensCrawled pages that are not auth-gated. 0 is a login-only site. Omitted when the run did not crawl.
engine_versionPackage version that wrote the run (ENGINE_VERSION).
statsFlow and bug counts. bugsCritical, bugsMajor, bugsMinor.
bugs[]Confirmed site bugs. minor does not fail the CLI.
signup_statusSet when a sign-up form was seen. Without your inbox: not_tested / no_inbox.
suggestionsSeparate from bugs. Missing autocomplete, viewport, or description, overflow on a page with no viewport, third-party page errors, and page errors with no script URL. Group by pageUrl. Not in the bug count or the verdict.
warningsModel errors, and a line when plan.plannerFallback replaced a failed planner call. A login wall that returns no journeys does not set the flag.

The field list is the report reference. Blocking bugs are every severity except minor. That is also what makes run exit 1.

Screenshots ​

FileWhen
.monkeytest/shots/page-*.pngOne viewport per crawled page
.monkeytest/shots/flow-*-step-*-*.pngOne PNG per step
.monkeytest/shots/flow-*.webmOne video per flow

StepResult.screenshotPath is an absolute path on the machine that ran the test. There is no .monkeytest/screenshots/ directory.

Next: Rerun and diff.

AGPL-3.0-only