Skip to main content

Getting Help

When something does not work as expected, msg.ZenTestAI gives you tools to investigate yourself and to share the relevant context with the support team in one click. This page covers both — the self-service triage path and the support-bundle workflow.

Channels​

The official support channel is e-mail:

You can also use the Feedback form on the bottom-left of the application to send a quick note from inside the product. The feedback form is the fastest way to drop a short observation; e-mail is the right channel for anything that needs a longer thread or attachments.

Before contacting support​

A surprising number of "the test is broken" reports turn out to be either a stale cache, an ambiguous step description, or a real application bug. Before opening a ticket, walk through the Failed-execution triage checklist:

  1. Read the Result and Explanation in the failing step's popover.
  2. Check the Warnings tab if there is one.
  3. Open the Log and scroll to the failing step.
  4. Decide whether the issue is cache, description, selector, or a genuine application bug.

If after that you still need help, capture a support bundle as described below — it cuts the support team's investigation time dramatically and usually means an answer arrives one round-trip earlier.

Collecting support information​

The recommended way to send context with a ticket is the support bundle flow:

  1. Open the failing execution's detail page.
  2. Click the Log button in the toolbar.
  3. In the Log popover toolbar, click Collect Support Information (bug icon).
  4. The dialog asks which categories of extra data to capture — see Categories of extra data below. Pick what's relevant and confirm.
  5. The platform re-runs the test in verbose mode. This may take a few minutes, especially if you selected HTML content or memory snapshots.
  6. When the verbose rerun finishes (passed or failed), the Download Support Information button replaces the Collect button in the Log popover toolbar. Click it to download a ZIP bundle.
  7. Attach the ZIP to your e-mail (or upload to your shared support folder if file size is an issue).
tip

Trigger the verbose rerun as soon as you can after a failure. Long delays between the original failure and the verbose rerun reduce the chance that the same problem reproduces. Application state, test data and external dependencies all drift over time.

note

The Collect Support Information button is only visible when the tenant has Detailed Service Request enabled and the administrator has granted you access. If you don't see it, ask your tenant administrator to enable it under Tenant Settings.

Categories of extra data​

When you trigger a verbose rerun, you can opt into four categories of extra capture:

  • AI communication — every prompt sent to an AI model and every response received. Almost always useful — most "why did the AI do that?" questions are answered by reading the actual prompt-response pairs.
  • HTTP communication — every network request the browser made, with request URL, method, headers, status, response body. Useful when the failure looks application-side rather than AI-side.
  • HTML content — full HTML snapshots of the page at each step. Useful when an element wasn't found and you want proof of whether it was actually in the DOM.
  • Memory snapshot — runtime memory state of the runner process. Useful for diagnosing runaway memory use or leaks.

Selecting more categories produces a bigger bundle and a longer rerun. If you're not sure, AI communication + HTML content is the most useful pair for everyday step-failure triage.

What the bundle contains​

A support bundle is a single ZIP that always contains:

  • The test case definition (current version).
  • The test execution record, with all resolved step descriptions.
  • The AI analysis (result + explanation) of every step.
  • All logs from the verbose rerun.

Depending on your tenant policy and the categories you selected when triggering the rerun, the bundle additionally contains:

  • The tenant settings (sanitized).
  • The application's configuration.
  • The captured AI communication logs.
  • The captured HTTP (API) communication logs.
  • The captured HTML content of the page under test.
  • The memory snapshot of the runner.

What is not in the bundle​

  • Secrets. Passwords, API keys, OAuth tokens, the values of fields explicitly marked as secret in your test parameters — none of these are included. The platform strips them before the bundle is assembled.
  • Anything from other tests or executions. The bundle is scoped to the one execution you triggered it from.
  • The original (non-verbose) execution. Only the verbose rerun is captured.

Encryption and sensitive data​

AI communication and HTTP communication can contain sensitive information about the application under test — page content, user data, business records — even though no secrets are included. Those two logs are therefore encrypted with msg-internal public-key encryption before being written to the bundle. Only the msg support team can decrypt them.

This means:

  • You can safely attach a support bundle to a normal e-mail without worrying that intermediate systems could read the sensitive parts.
  • You yourself cannot read the encrypted log files in the bundle — they are intended for support.
  • Bundles can therefore safely be archived or shared internally for record-keeping without exposing application data.

What to include in the ticket itself​

Even with a support bundle attached, a short written description helps support route the ticket and prioritize:

  • What were you trying to do? ("Record a test for the Order page.")
  • What happened that you didn't expect? ("Step 7 fails with 'element not found' but the button is clearly visible in the screenshot.")
  • What have you already tried? ("Cleared cache for the step, rewrote the description, neither helped.")
  • A link to the execution. Copy the URL of the detail page; the platform team can open it directly and reproduce the failure.

The first three points map roughly to "expected", "actual" and "tried" — the universal trio of any bug report.

Common topics support can help with​

  • Understanding a test failure — the bundle + a ticket is usually enough for the team to explain whether it was a cache issue, a prompt issue, or a real application defect.
  • Provisioning a new AI model or vendor — see AI-Models for the supported list; for anything not in that list, contact us at hello@zentest.ai.
  • Tenant-level policies — parallel-execution limits, detailed-service-request opt-in, BYOK key rotation, e-mail notifications.
  • External-system connectors — Jira XRay / Zephyr token issues, status mapping not behaving as expected, sync failures (see External Systems).
  • Pipeline integration — the platform's REST API for CI/CD use (Pipeline Integration).

Self-service references​

A lot of "support" questions are already answered in the documentation. Common places to look first: