Troubleshooting & FAQ
Updated Jul 2026

Documentation › Troubleshooting & FAQ

Troubleshooting & FAQ

User  The problems that come up most, and how to resolve them.

On this page

Understanding run states & results

A run has a state (is it finished?) and a result (did it pass?). The states you'll see in the Runs list:

StateMeaning
Active (spinner)Running now.
SuccessPassed.
FailedAn assertion failed — the test ran but a check didn't hold.
BrokenA non-assertion error (e.g. an element wasn't found, a script error).
StoppedManually stopped.
QueuedWaiting for a free execution node (free plans have no parallelism).
WaitingWaiting on a dependency to start.
Failed to startThe run couldn't be launched at all — rare; contact support if it persists.

Failed vs Broken: "Failed" means your test found a problem (an assertion) — often a real defect. "Broken" means the test itself hit an error (a bad locator, a timing issue, an environment problem) — usually a test to fix, not a product bug.

Run problems

SymptomLikely cause & fix
Run stuck in QueuedNo free node. On a free plan runs are serial — wait, or upgrade for a private pool. Check the Runs x/y counter in the status bar.
Run is Broken with "element not found"The locator is stale or you're in the wrong frame. Re-record the step, use a more stable locator, or add a waitForElementPresent. For Maximo, confirm you're inside #manage-shell_Iframe — see the Maximo playbook.
Run hits the time limitmx.run.time_to_live caps run duration (default 300s). Increase it (up to the plan limit) or split the test.
No live videoLive video appears a few seconds in and only for single-case runs. Video recording is separate and needs enable_video_recording=true.
MXML case won't compileCheck for an unsupported <ui:className> locator (replace with id/css/xpath). View the compile log via Workspace ▸ More options ▸ Compile logs.

AI test cases

SymptomLikely cause & fix
Run failed: "did not report a result"The agent never reached a conclusion. End your script with an explicit "Report what was checked" step, and make sure earlier steps aren't ambiguous or looping.
Failed with "loop detected"The agent got stuck repeating an action. Make the step clearer/more specific, or adjust mx.playwright.aggressiveness_level.
The AI did the wrong thingOpen the run's Console output to see each step the agent took, then tighten the wording of the offending instruction (name the exact button/field).
No MXML case was producedRecording needs both mx.record_maximo=true and the recorder credentials (mx.recorder.project_name + mx.recorder.maxtaf_api_key). Missing either → the run executes but records nothing.
AI Assistant / AI case errors about the providerCheck Project Settings ▸ AI Configuration — the provider key/model must be set for the provider you're using.

Recorder & upload

SymptomLikely cause & fix
Case exports but doesn't appear in MaxTAFCloud upload requires PROJECT_NAME, BASE_URL and MAXTAF_API_KEY set, and SAVE_TO_FILE_ONLY=false. Check the recorder's Configuration and validate the API key.
Upload rejected (auth error)The API key or project id is wrong, or the key was invalidated. Regenerate it in Admin ▸ User Settings and re-enter it in the recorder.
Case name rejectedNames must contain only letters, numbers, spaces and underscores. Remove other characters.
Clicks recorded with a CSS selector instead of an idThe element had no usable id at capture time. Add an assert/store on the field via right-click to capture the id, or edit the locator in the generated MXML.
Nothing records inside MaximoThe extension must be allowed to run in all frames. Reload the Maximo tab after enabling the extension so the content script injects into the iframes.

Parameters

SymptomLikely cause & fix
A parameter value isn't taking effectLayers combine by mx.params.source / mx.params.method. A case-level value overrides a project-level one; an empty case-level value can blank a project value. Check the case's parameter config.
Can't create an mx.* parametermx.* names are reserved/standard — set their values, but use custom names for your own variables.
Wrong environment being testedConfirm mx.maximo.address (and any environment URL params) point where you expect; override per run with runParams from CI.

General FAQ

What's the difference between the AI Assistant and an AI test case?

The AI Assistant is a chat helper that writes/debugs code for you. An AI test case is a runnable test the agent executes. See Writing AI Test Cases.

Which case type should I use?

For Maximo, MXML (usually via the recorder or an AI recording run) is the fast, token-free choice for regression. Use an AI case to author quickly or explore. Java/Python/JavaScript are there when you need full programmatic control.

Where are the AI execution logs?

In the run's Console output section — not the "Log" section.

How do I back up my work?

Use project Export (Settings ▸ Main) for cases/suites/schedules/files/params. Runs aren't included — download the runs folder separately. See Backup & maintenance.

Can I run tests from CI?

Yes — see CI/CD Integration.

How do I get more help?

Use Learn ▸ Support in-app, or the in-app AI Assistant for authoring questions.