Automations
Rules that react to an event, check their conditions and run actions in the background, with run history, retries, an emergency stop and loop limits.
Automations are rules that react to something happening in UnitX: “when this happens, do that”. Each rule listens for one kind of event, checks its conditions and runs its actions once, in the background, just after the change that set it off has been saved. Use them for small follow-ups that otherwise depend on someone remembering, like tagging urgent work, assigning the person who handles a queue, or asking someone to approve.
Where to find it#
Open Automations from the sidebar. The page has three tabs:
- Rules: create rules, switch them on and off, copy and delete them.
- Run history: every time a rule ran, including the runs it skipped and why, and the result of each action.
- Settings: the organization-wide emergency stop and the limits that prevent loops.
Key concepts#
- Rule
- A named automation made of a trigger, optional conditions and one or more actions. A rule is either On or Off.
- Trigger (“When”)
- The event the rule listens for, such as “A work item is created” or “A form is submitted”. A rule listens for exactly one event type.
- Conditions
- Checks that decide whether the actions run at all. A rule can also be limited to changes that touch particular fields; its summary then reads “only when … changes”.
- Action (“Then”)
- What the rule does: add a tag, assign someone, send a notification, raise an approval, route a form submission, and so on.
- Run
- One rule reacting to one event. Every run is recorded in Run history with a status and the outcome of each action.
- Chain depth
- Actions can themselves cause events, so one automation can set off another. The depth counts how many automations deep a chain has gone.
Automations or workflows?
Who can use it#
Permissions
automation.viewautomation.manageautomation.view lets you see the rules, their run history and the settings. Creating, switching, copying and deleting rules, trying a failed run again, and changing the settings all need automation.manage. The default Member role holds both; the Viewer role can only look.A rule acts with the authority of the person who created it. Some actions need a permission of their own (for example work_item.manage to set a priority). The rule builder shows that requirement under the action, and the author must still hold the permission each time the rule runs. If their role changes or they leave, the actions that need the lost permission stop working. Have someone with the right permission copy the rule, which makes them its author.
Create a rule#
Open the Rules tab
Go to AutomationsRules. The form for a new rule is at the top of the tab. You only see it if you have
automation.manage.Name the rule
Type a Name that says what the rule does, for example “Tag urgent work”. The name can be up to 200 characters.
Choose when it runs
Pick the event from When. The available triggers are listed in the trigger reference below.
Choose what it does
Pick an action from Then. A short description of the action appears underneath, along with the permission it needs, if any.
Fill in what the action needs
A box appears for each setting the action requires, labelled with the setting’s name, such as
tagIdorrecipientId. Settings that refer to a tag, person, project or approver take that record’s ID. The Create automation button stays disabled until every required box is filled in.Decide what happens if an action fails
Under If an action fails, choose Carry on with the rest (the default) or Stop, and skip the rest. Rules set to stop show a Stops on failure badge in the list.
Create it
Select Create automation. The rule is saved and switched on straight away. If something about it is invalid, such as a missing setting or an action that does not exist, UnitX refuses to save it and the message says which part is wrong.
Tip
What the builder covers
Switch, copy and delete rules#
Each rule in the list shows its name, an On or Off badge, and a one-line summary of its trigger and actions. People with automation.manage also see these controls:
- The switch turns the rule on or off. The change takes effect straight away and nothing is deleted.
- Copy creates a duplicate named “… (copy)”. The copy always starts switched off, so it never doubles up the original’s effect by accident, and you become its author.
- Delete removes the rule permanently, together with its run history. If you only want it to stop, switch it off instead.
You can’t edit a rule’s trigger or actions from this page. To change what a rule does, create a new rule and then switch off or delete the old one.
Check what a rule did#
Open Run history
Go to AutomationsRun history. The newest runs are at the top.
Filter by outcome
Use Show to narrow the list: Everything, Failed, Will retry, Done, Skipped — conditions or Stopped — too deep.
Open the detail
Select Detail on a run to see each action on its own: whether it succeeded, how many times it was tried, any error, and the input it was given and the output it produced. If no actions are listed, the run was skipped before it got that far.
Try a failed run again
Fix the cause first, for example a deleted tag or a person who left. Then select Try again on a run that is Failed or Will retry. Actions that already worked are not repeated, so trying again finishes the job rather than doing half of it twice.
Run statuses#
| Status | What it means |
|---|---|
| Running | The rule is working through its actions now. |
| Done | Every action ran. Some may report that there was nothing to do, for example a tag that was already there. |
| Failed | At least one action failed for a reason that retrying will not fix on its own. The error is shown on the run. |
| Will retry | A temporary problem stopped the run. UnitX tries it again automatically, and you can also select Try again. |
| Skipped — conditions | The event matched the trigger, but the rule’s conditions were not true, so no action ran. |
| Skipped — field | The rule only reacts to changes to certain fields, and this change did not touch any of them. |
| Stopped — too deep | The run was part of a chain of automations that reached the depth limit, so it stopped instead of looping. |
A run started by another automation says “set off by another automation (depth N)”. A run tried more than once shows how many attempts it took.
Emergency stop and limits#
Emergency stop#
AutomationsSettingsEmergency stop pauses every automation in your organization at once. The badge changes from Running to Paused. Nothing is deleted and each rule keeps its On or Off setting.
Warning
Limits#
Two limits protect your organization from runaway rules. Enter a new value and select Save limits.
| Setting | Default | Allowed values | What happens at the limit |
|---|---|---|---|
| How many automations deep | 10 | 1–100 | A chain of automations that reaches this depth stops with the status Stopped — too deep. |
| Runs per minute | 300 | 1–100,000 | Extra events wait their turn and run once the burst has passed. They are delayed, not dropped. |
Raise the depth limit only if you are sure your rules can’t trigger each other in a circle. Your subscription plan may also limit how many automation runs you can use; see Billing and plans.
Trigger reference#
| When | Runs when |
|---|---|
| A work item is created | Someone creates a work item. |
| A work item is edited | Any field on a work item is saved. |
| A work item’s priority changes | The priority is changed. |
| A work item is assigned | The assignee is set or changed. |
| A work item’s status changes | The item moves to another status. |
| A work item is tagged | A tag is added. |
| Someone comments | A comment is posted. |
| A form is submitted | A new form submission is received. |
| A request is raised | A form submission becomes a request. |
Action reference#
The Then list shows each action by name. Actions that work on “the work item” or “the request” use the record the event is about, so pair them with a matching trigger.
| Then | What it does | Required settings | Author needs |
|---|---|---|---|
| add tag | Adds an existing tag to the work item. | tagId | work_item.manage |
| assign user | Assigns the work item to a specific member. | userId | work_item.manage |
| set priority | Sets the work item’s priority. | priority | work_item.manage |
| send notification | Sends an in-app notification to a member. | recipientId, title | Nothing extra |
| request approval | Raises a single-approver approval request for this run. | approverId | approval.manage |
| route submission | Turns a form submission into a request with a type, priority and initial assignment. | requestType | form.manage |
| request approval for request | Raises a multi-approver approval on the request. | approverSource | approval.manage |
| convert request to work item | Creates a work item from the request, with a title and description built from the submitted answers. | projectId, typeKey, titleTemplate | work_item.create |
Route form submissions into requests#
A form submission doesn’t become a request by itself. Routing is done with automation rules, so your intake queue fills itself the same way every time.
Route each submission
Create a rule with When set to “A form is submitted” and Then set to “route submission”. Enter the
requestType, for examplepurchase_request. Each submission then becomes a request of that type in Requests. Routing the same submission twice never creates a second request.Choose how it is assigned
Routing can leave the request unassigned, give it to one named person or to a team, rotate round-robin through a team’s members, or pick the team member with the fewest open requests. The rule builder doesn’t show these optional settings, so a rule made there puts requests in the unassigned queue (Nobody yet on the Requests page), where anyone with
form.managecan assign them.Ask for an approval, if needed
A rule triggered by “A request is raised” can use “request approval for request”. Set
approverSourcetospecific_users,manager,roleorteam. The approval can require everyone, any one, a minimum number, or a percentage of approvers (50% is a majority). A request never gets more than one approval from the same rule. The request then shows what it is waiting on, and the approval links back to the request.Turn requests into work
Another rule on “A request is raised” can use “convert request to work item”. The
titleTemplateis text with placeholders:{{subject}}is replaced by the answer to the field whose key issubject, and{{reference}}by the submission’s reference number, such as REQ-42. A request priority of high becomes urgent on the work item, so urgent requests don’t lose their urgency.
Public form submissions
Tips and limitations#
- Rules run just after a change is saved, not during it. A failing automation never undoes your edit, and it usually runs in under a second.
- Start new rules with a narrow trigger. “A work item’s priority changes” runs far less often than “A work item is edited”.
- Use Copy before a risky change. The copy starts off, so you can switch it on when you are ready and switch the original off.
- Two rules on the same form with different conditions are how you get two intake queues.
Troubleshooting#
Why hasn’t my rule done anything yet?
Check that the rule is On, that SettingsEmergency stop isn’t paused, and that the event really matches the trigger. Then open Run history. A skipped run tells you whether the conditions or a field filter stopped it.
Why did a run say “Skipped — field”?
The rule only reacts when certain fields change, and the change that reached it didn’t touch any of them. It is recorded so that “why didn’t my rule fire” has an answer.
An action says the author doesn’t hold a permission.
A rule can only do what its author can still do. If their role changed, have someone with the right permission copy the rule, switch the copy on and delete the original.
A placeholder came out blank.
The field it named had no answer, or no field has that key. The run’s detail lists which placeholders found nothing, which is usually a typo in the field key.
A routed submission says “failed”.
Something the rule refers to couldn’t be used, often a project, work item type or team that no longer exists, or a submission from a public form. Nothing is thrown away. Fix the rule, then use Try again on the failed run.
Can one automation set off another?
Yes. That’s why there is a depth limit. When a chain reaches it, the run stops and says so instead of looping.
Related pages#
Still stuck? Search the docs with ⌘K, open Help inside UnitX, or contact support.