For the complete documentation index, see llms.txt. This page is also available as Markdown.

Reading a Report

Every run produces a report — per-step screenshots, timestamped logs, a video recording and DT usage.

Every run produces a report: what each step did, what the screen looked like before and after, what Drizz decided, and what it cost.

Platforms

Android · iOS

Every report has

Per-step status · before/after screenshots · timestamped logs · DT usage

Three types

Standard · Debug · Accessibility

Links

Signed and time-limited — they expire

Prerequisites

  • A completed run, local or cloud

  • Access to the project the run belongs to

What's in every report

Element
Contents

Per-step status

Passed, failed, healed or skipped, for every step in order

Before and after screenshots

The screen going into the step, and coming out of it

Timestamped logs

What Drizz decided at each step and why, with the duration

A video recording

The whole run, downloadable from the Actions section — see Recordings

Token usage

DT spent on the run

The device that ran it

Model and OS version

Steps that resolved from cache are marked "Predicted Action drizzing fast ⚡" — see Caching. Steps repaired mid-run carry a healed badge — see Self-healing.

Read a report

  1. Read the run status: Passed, Passed (healed), Failed, Error or Blocked. Error and Blocked are not test failures — see Statuses.

  2. Read the healed-step count, including on a green run. Healed steps have weak descriptions.

  3. Open the first failing step, not the last. A cascade of failed steps has one cause at the top.

  4. Open that step's before-screenshot and compare it against the screen the step expected — see When a step fails.

  5. Open the video only if the stills don't explain the failure. Use it for timing, animation and navigation problems.

Report types

The run, step by step, with before and after screenshots, timestamped logs, the recording and DT usage.

This is the default report and is produced for every run.

Adds deeper diagnostic detail beyond the standard report.

Applies when the standard report does not explain a failure.

Adds accessibility findings from the run, alongside the functional result.

Drizz validates what is visually rendered. It does not do WCAG conformance scanning, screen reader testing, contrast analysis, or semantic and role-level validation — see Known limitations.

Sharing a report

Report links are signed and time-limited. They open for anyone holding the link until the link expires, then stop.

Use
Supported

Share for immediate review — a message, a live triage call

Attach to a ticket or a release record

Link

Getting results out

There is no export format. Reports cannot be exported as JUnit XML, JSON or Allure, so CI systems cannot ingest results directly. With no run-status endpoint, a pipeline can trigger a Drizz run but cannot wait on it or gate a build on the outcome. Artifacts download by hand from the Actions section.

Common mistakes

What you do
What happens

Debug the last red step

It's a consequence. The first failing step is the cause

Read Passed (healed) as Passed

The suite is held together by repairs and the pass rate hides it

Paste a report link into a ticket as the evidence

The link expires and the ticket becomes unusable. Attach the artifacts

Open the debug report first

The standard report's before-screenshot answers most failures faster

Treat Error as a test failure

It's infrastructure, not your test. Re-run it before investigating

Expect CI to parse the report

No export format exists. There is nothing to parse

Next


Last updated: 6 August 2026

Last updated

Was this helpful?