Using Custom Tasks
A custom task works on form data without anyone starting it: it reacts to a manual run, a schedule, or a record change, and then processes data, asks someone to handle a form task, calls an agent, calls a script or integration flow, sends messages or visualization cards, and can wait for external callbacks.
Tasks belong to the current workspace. Open Work, then Automation Tasks.
Before You Create A Task
- You need the form the task works on; data processing, human processing, branch, and recipient fields are read from the bound form
- Scripts and integration flows must be published before a service call node can select them
- Message nodes need an active message channel and template; visualization nodes need a report or chart that already exists
Task List
The list supports keyword search and creation, plus opening the design canvas, version history, running the task, execution history, activation, pausing, and deletion.
| Column | Description |
|---|---|
| Name | Task name |
| Code | Task code |
| Trigger | Manual, schedule, or data change |
| Status | Draft, running, disabled, or archived |
| Schedule sync | Shows when the schedule configuration is syncing or failed to sync |
Task Status
| Status | Description |
|---|---|
| Draft | The initial state of a new task; it does not react to its trigger |
| Running | Published and active, so it runs according to its trigger |
| Disabled | Stops the trigger from starting new executions; running executions continue |
| Archived | The task is archived and is no longer used as a regular task |
Create A Task
- Click Create automation task
- Enter the Name and an optional Description
- Click Create
- The new task is a draft; click Open design canvas to configure the trigger and nodes
- After saving and publishing a version, the task runs according to its trigger
Trigger settings, record scope, and node configuration are covered in Automation Task Design Canvas.
Trigger Methods
| Trigger | Description |
|---|---|
| Manual | Started from the task list with Run now after selecting a record |
| Schedule | Runs on a Cron expression or a fixed interval |
| Data change | Starts when a form record performs the configured action, optionally limited to selected fields |
A data-change task takes its form and execution target from the trigger itself, so it does not configure a separate record scope.
Run, Activate, And Disable
| Action | Effect |
|---|---|
| Run now | After searching and selecting execution data by record identifier, queues this run |
| Activate | Lets the task react to its trigger again |
| Disable | Stops the trigger from starting new executions; running executions continue |
| Delete | Permanently deletes the task |
When the same task is already running and its overlap policy is "skip while running", the new run request is skipped.
Version History
Version history lists every saved version with its status (draft, published, or archived), the creation time, and the operator. View opens that version in the designer, and Publish makes a draft or archived version the active one.
Verify A Task
- Save and publish the version, then use Run now and pick a record
- Open Execution history and check that a new execution appears
- Read its status, current node, and progress; a successful execution ends with the status succeeded
Execution fields and details are covered in Automation Task Execution History.
Troubleshooting
| Symptom | What to check |
|---|---|
| The schedule never runs | The task must be published and active; a failed schedule sync means the schedule was not registered |
| A record change does not start the task | The data-change trigger needs the matching form and action, and the listened fields must contain the field that changed |
| The run request reports that it was skipped | The previous execution is still running and the overlap policy skips concurrent runs |
| No execution record appears | Executions are only created when the task actually starts; check the trigger and the task status |