- Release notes
- Overview
- Getting started
- Frequently Asked Questions
- Frequently asked questions
Healing Agent user guide
Frequently asked questions
Healing Agent is currently available in Automation Cloud only.
Healing Agent is available in Windows and cross-platform project types.
Healing Agent is only available for UI Automation Modern activities.
Healing Agent is currently only available for processes executed as jobs. Healing Agent is not available for other types of jobs, such as triggers or test cases.
Healing Agent is available both for desktop and web applications.
In addition, Healing Agent can recover automations designed to run on remote environments, such as Citrix.
Healing Agent is most effective for UI-based automations in dynamic and unpredictable environments. This includes automations in unstable or frequently changing applications, such as cloud-based platforms, where updates are common.
Healing Agent is also particularly useful in environments that are not tightly controlled and which are therefore subject to operating system updates, resource contention and other factors of inconsistency that impact automation reliability.
Some Healing Agent recovery strategies can be applied to existing projects, as long as you update the UI Automation package to the minimum required level.
For more information, refer to Prerequisites.
Yes, Healing Agent suggests updates for elements in the Object repository. However, Healing Agent has different behaviors depending on whether it's operating at runtime or design time:
- At runtime: Healing Agent suggests fixes and performs self-healing regardless of whether the target element is part of the current project or a referenced dependency library.
- At design time: Healing Agent can only apply fixes to elements that are part of the same project. If the element belongs to a dependency library, you must open that library project, import the debug file, and apply the fix there.
PUT /odata/Releases({key}).
Healing Agent is configured in the ProcessSettingsDto section,
through the AutopilotForRobots properties.
Healing Agent uses two primary safety mechanisms to prevent the selection of incorrect selectors:
- Administrator control:
Administrators can configure Healing Agent to provide fix recommendations
without automatically implementing self-healing actions.
Administrator control allows developers to review proposed selectors and manually fix issues, ensuring accuracy and control.
- Multi-strategy voting
system: When Healing Agent is permitted to self-heal and a selector is
not found, it executes multiple recovery strategies. Each strategy proposes a
candidate selector, and the candidate with the most votes is considered eligible
for use.
The voting algorithm minimizes the risk of selecting an incorrect selector.
These mechanisms work together to balance automation efficiency with the need for accuracy and control in the self-healing process.
- Is Healing Agent available both in cloud and on-premises deployments?
- What project types is Healing Agent available for?
- Is Healing Agent available both for UI Automation Modern and Classic activities?
- What types of jobs is Healing Agent available for?
- What types of applications or environments is Healing Agent available in?
- What types of automations are most suitable for Healing Agent?
- Is Healing Agent available for existing projects, or only new ones?
- Does Healing Agent recognize and suggest updates for elements in the Object Repository?
- Can Healing Agent be enabled or disabled at scale?
- What safety mechanisms prevent Healing Agent from choosing the wrong selector?
