How to Use {{agentResponse}} in Atlassian Automation: Unlocking the Power of AI-Driven Workflows

[#1]

Atlassian’s Rovo Agent brings AI-powered automation to your Jira and Confluence workflows. But to truly harness its potential, it’s essential to understand how to use the smart value {{agentResponse}}—the key to capturing and leveraging the agent’s output in your automation rules.

In this post, we’ll break down what {{agentResponse}} is, how it works, and practical ways you can use it to supercharge your team’s productivity.

---

What is {{agentResponse}}?

When you invoke the Rovo Agent in an automation rule, it generates a response based on your prompt—whether that’s analyzing a form, researching a trend, or summarizing content. The smart value {{agentResponse}} lets you capture this output and use it in subsequent automation actions.

Because Atlassian automation doesn’t allow agents to take actions autonomously, the real power comes from using the agent’s response to drive updates, create content, or trigger further workflows.

---

Real-World Use Cases

Here are some of the most popular ways teams are using {{agentResponse}}:

1. Update Issues with AI Insights

   • Example: The agent analyzes a submitted form and calculates a score based on knowledge sources. Automation then updates a custom field in the issue with this score.

2. Create or Update Confluence Pages

   • Example: The agent researches an upcoming trend. Automation creates a Confluence page listing all the ideas in a formatted table.

3. Parse JSON and Trigger APIs

   • Example: The agent returns a list of ideas in JSON. Automation parses the JSON and creates Jira Product Discovery (JPD) issues for each idea.

---

Understanding the Flavors of {{agentResponse}}

The {{agentResponse}} smart value comes in several formats, each suited to different automation actions:

Format	Notation	Description	When to Use
Markdown	{{agentResponse}} or {{agentResponse.markdown}}	Default format, great for rich text in Confluence and Jira’s new editor.	Use for most content, especially in Confluence.
ADF (Atlassian Document Format)	{{agentResponse.adf}}	Structured JSON for rich content across Atlassian products.	Use where ADF is supported.
Wiki	{{agentResponse.wiki}}	Legacy Jira markup, phased out in favor of ADF.	Used automatically for some Jira actions.
JSON	{{agentResponse.asObject}}	Raw JSON for advanced parsing and automation.	Use when you need to extract specific values or trigger APIs.
JSON	{{agentResponse.asList}}	Raw JSON for advanced parsing and automation.	Similar as asObject, but you can branch on the results immediately in subsequent steps.
String	{{agentResponse.asString}}	Plain string for further manipulation (split, concat, etc.).	Use for string operations in automation rules.

Tip: The system often selects the right format for you, but you can specify the format explicitly if needed.

---

Best Practices & Tips

• Valid JSON Only: If you want to parse the agent’s response as JSON, ensure the agent returns valid JSON—no extra code blocks or comments.

• Multi-Shot Prompting: Train your agent with clear, multi-step prompts to get structured, predictable outputs.

• String Manipulation: Use .asString for splitting lists or further processing in automation.

• Automation Compatibility: JSON is best supported for parsing in Atlassian Automation today.

---

Frequently Asked Questions

Q: What happens if I use {{agentResponse}} without specifying a format?
A: The default is Markdown, which works well for most content. For structured data, use .asObject or .asString.

Q: Can I use YAML instead of JSON?
A: Automation currently supports JSON parsing natively. YAML may be useful for future agent-to-agent workflows, but stick to JSON for now.

Q: Which automation actions support {{agentResponse}}?
A: Most actions that update issues, create pages, or trigger webhooks can use {{agentResponse}}. The format you use may depend on the action.

---

By mastering {{agentResponse}}, you can unlock powerful, AI-driven automation in Atlassian—saving time, reducing manual work, and letting your team focus on what matters most.

Ready to get started? Try adding a Rovo Agent action to your next automation rule and see the possibilities for yourself!

[Janessa Drainville]

'Can't wait to give this a whirl, @Jensen Fleming! 🙌 

[Alex Gallien]

Thanks for getting these published @Jensen Fleming! We've been making heavy use of {{agentResponse.asObject}} to combine agents with automation and it's a powerful tool.

For anyone who wants to see an example of this in action, check out our Let's build a Rovo Agent workshop series. In particular the JSM Triage agent we featured last month is available on demand and contains an example prompt and automation rule configuration (click the handouts button once loaded).

[Josh]

@Jensen Fleming thanks for the super helpful article!

Is there a way to differentiate between {{agentResponse}} values when multiple agents are included in an automation rule (e.g. branched executions or sequential calls from one agent to the next one)?

[Rebekka Heilmann _viadee_]

Thank you for this article, @Jensen Fleming !

You can also do other generic Smart Value actions with the AgentResponse which can be super helpful.

[Jensen Fleming]

@Josh AgentResponse will use the value from the result of the previous node. If you want to save that response for later, you can save it as variable via `Create Variable` action.

[Brandon Davies]

Thanks for this great article! I have been using the JSON output and will check out the above resources from @Alex Gallien to see how I can make it more consistent as it still gets some infrequent hiccups around JSON formatting.

[Alexander Ziegltrum]

@Jensen Fleming I joined the Workshop today, where an automation uses the "Dexter - Triage Automator" Rovo agent.

In the automation these smartvalues must be used, as in the handout of the workshop: 

• {{agentResponse.asObject.priority}}
• {{agentResponse.asObject.region}}
• {{agentResponse.asObject.team}}

But the automation reports an error in the audit log:

Unable to render smart values when executing this rule:
Failed to get value for agentResponse.asObject.team: Team: {{agentResponse.asObject.team}}

Which leads to that the values can't be used for editing and updating the issue as intended with the automation.

It is the same with:

• {{agentResponse.asList}}

Is this to be expected or what is the problem with that?

Thanks for a hint.

Best regards, Alex

[Alex Gallien]

@Alexander Ziegltrum my guess is that the agent is not correctly returning the Team value in the JSON response. A couple things to try to validate that:

• Open up your agent in chat and prompt it with AUTOMATION Analyze <valid issue key>. Does the expected JSON object get returned? See an example in the instructions for the Automation scenario.
• Another approach would be to add a log action to your automation rule that just logs {{agentResponse}} after the Rovo agent is called.

Some other things to think about:

• If you're following that workshop directly, did you create the Team definitions page as listed under Demo Data? That is what drives the team assignment.
• I realize on reviewing that handout that there is a direct link in the prompt to the team definitions page from my test environment - updated the instructions accordingly to overwrite that with a link to the doc in your environment.