Auto-Recovery

How It Works

A test runs and a step fails (e.g., a button can’t be found)
The CLI’s AI analyzes the current page state — what’s visible, what changed
It attempts to recover: finding the element by alternative means, waiting for the right condition, or adapting to the new UI
If recovery succeeds, the test continues and the step is marked as healed
If recovery fails, the test is marked as failed and escalated to auto-healing

This all happens transparently during test execution — no manual intervention needed.

Issue	How It Recovers
Changed selectors	Finds the element using alternative attributes (text, role, position)
Timing issues	Waits for the right conditions before proceeding
Minor UI changes	Adapts to layout or text changes that don’t affect the flow

Some issues require deeper analysis and code changes:

These are handled by Auto-Healing, which creates a full AI agent session to fix the tests.

In your test run results, recovered tests show up in the healed count:

Result	Meaning
Passed	Test passed without any recovery needed
Healed	Test had a failure that was automatically recovered
Failed	Test failed and couldn’t be recovered

Auto-recovery runs when your project’s run mode is set to Auto-Heal. This is controlled by the runMode option in checksum.config.ts:

Value	Behavior
`RunMode.Normal`	Tests run normally and fail on the first error — no recovery (default)
`RunMode.Heal`	Failing steps automatically attempt recovery before being reported as failures

Set runMode: RunMode.Heal in checksum.config.ts to enable it. See the runMode reference for full details.