Automating Rovo agents
Admins can use agents in automation rules to save time on more complicated repetitive tasks. Using an agent within an automation rule allows you to generate and share content in any app integrated with Atlassian Automation. There are two ways to this: add a trigger to an agent’s configuration, or add the agent to an automation rule.
Add a trigger to an agent
If you are an agent’s administrator or collaborator, you can add a trigger in the agent’s configuration.
Use the app switcher to navigate to the Studio app
Select Agents
Browse to locate the agent you want to automate
Select the agent to open it
Select Triggers from the left-hand navigation
Any existing rules that trigger this agent are listed, organized by app
Select Add automation trigger to configure an event that will trigger this agent
Select the appropriate app and space
You’ll be redirected to automation in Jira, Jira Service Management, or Confluence to finish building the rule
Save and enable your rule
Refresh your agent’s Triggers screen to see it listed
Add an agent to an automation rule
If you are a space or app administrator, you can create an automation rule that invokes an agent to perform a secondary action.
Use the app switcher to navigate to the Studio app
Select Automation
Select the app and space you want to create the automation rule in
Select Get started
You’ll be redirected to Automation in the app and space you selected
Select Create rule > Create from scratch
Add a trigger to your rule
Configure it if needed
Select Next > THEN: Add an action
Find and select the action, Use Rovo agent
Connect Rovo to automation to allow agents to act on your behalf
Select an agent (required)
Write a prompt (required)
Add a second action to tell the agent what to do
Configure it with
{{agentResponse}}(more details below)
Tips to configure the action, Use Rovo agent
Mention the trigger’s object in your prompt
When you use a Rovo agent in an automation rule, you’ll need to write a default prompt — in the same way you’d prompt an agent directly. Each time the rule is triggered, this prompt is sent to the agent on your behalf.
|
As long as the object in your prompt matches the object from your chosen trigger, the appropriate context will be provided.
If the trigger is Work item created, your prompt in Use Rovo agent might be:
Read and review the quality of the description field for this work item.
In this example, when the rule (and the agent) is triggered, the URL of the relevant work item is provided to the agent as context. The agent would analyze the quality of the work item’s description field and generate a response.
Add a second action – with text output
To use the agent’s response, you must add a second action that can leverage it. That is, it’s not enough to add a Rovo agent to a rule — you have to give it another action to execute.
This means if you want the agent to add its response as a comment, you have to add the Add comment action — whether or not an agent’s profile contains the Add comment skill.
And since the agent’s response is text-based, you must choose actions that generate text output. Think of it as if you were chatting with the agent — the response would appear as text in the Chat window.
For now, only a few actions generate text output (and can therefore use the agent’s response), but more will be added over time.
Use the smart value, {{agentResponse}}
Once you select a compatible secondary action, type the smart value {{agentResponse}} in the configuration field. This inserts a uniquely generated response each time the rule runs.
Smart values are dynamic variables that can make your rule more flexible. Each smart value is written as a hierarchy, starting with a top-level object and followed by properties of that object, {{object.property.subProperty}}.
Formatting of the {{agentResponse}} defaults to rich text markdown, but can be changed by adding properties.
Automated agents assume your app permissions since they operate on behalf of the rule creator. So for example when the {{agentResponse}} generated by the first action, Use rovo agent, becomes the comment that is added by the second action, Add comment — the agent only adds comments to pages you have access to.
The comment is attributed to the agent by default, but if you prefer you can reassign the author. For example, you can configure the Add comment action so the comment is attributed to you (the rule creator) rather than the agent. A note will still indicate that an agent auto-generated it.
Rule examples
See all available out-of-the-box agents
Scenario | Rule |
|---|---|
Use the agent, Decision director The automation rule is triggered when new Confluence pages are published with “DACI” in the title. The Decision director agent should review the page and leave a comment suggesting improvements. |
IF: Rovo condition Content in the title relates to given text: DACI (field text) THEN: Use Rovo agent, Decision director Review the content on this page and give recommendations (prompt) THEN: Add comment to page {{agentResponse}} (field text) |
Use the agent, Jira work item organizer The automation rule is triggered when a work item is created without a parent task or epic. The Jira work item organizer agent should find a relevant parent and leave the suggestion as a comment on the new work item. |
IF: Work item field condition Type is one of [all child work item types] AND IF: Related work item condition Parent does not exist THEN: Use Rovo agent, Jira work item organizer Find a relevant parent (or epic) for this work item and explain why you recommend it (prompt) THEN: Add comment to work item {{agentResponse}} (field text) |
Use the agent, Jira theme analyzer The automation rule is triggered every 2 days at 8am. The Jira theme analyzer agent should analyze data from work items created in the past 2 days and send a summary in Slack, organized by theme. |
Every 2 days at 8:00am THEN: Use Rovo agent, Jira theme analyzer Identify themes from work items created in the past 2 days (prompt) THEN: Send Slack message {{agentResponse}} (field text) |
Was this helpful?