Best practice for Rovo agent to create ticket comments via automation rules - Link embedding issue

[#1]

Hello everyone,

I'm currently using a Rovo AI agent within a Jira Cloud automation rule to provide onboarding assistance for access requests.

Current implementation:

• Automation rule triggers the Rovo agent with ticket data
• In the automation rule, I use "And: Add comment to work item" action
• The comment content is set to {{agentResponse}}

Issue encountered: During testing, I've found that I cannot embed clickable links within the {{agentResponse}} when it's added as a comment. I've tried both formats:

• Markdown format: [text](URL)
• Jira format: [text|URL]

However, the links appear as plain text instead of being rendered as clickable links in the ticket comments.

Questions:

1. Is there a more efficient or recommended way for a Rovo agent to directly create comments on Jira tickets through automation rules?
2. How can I make the Rovo agent response include properly formatted, clickable links in Jira comments?
3. Are there specific formatting requirements for links in {{agentResponse}} variables?

Any guidance or documentation you can point me to would be greatly appreciated.

Thanks in advance!

[Rimantas Andrulevičius]

@Jordi Romera could you elaborate on the use case a bit more (I'm also in a similar case creating an access mng solution on JSM and having the gray area of onboarding how-to)?

I don't have an exact answer for your current case, but what I've seen is:

1. Try without any format (Jira should handle plain links and render them based on user settings)

2. Since you are using automation, I would suggest trying to use JSON format and make a smart value ->variable (or use advanced branching) out of the response:

https://youtu.be/BmjCyM--etk?si=br8M8cms644efYlW&t=1351

https://community.atlassian.com/forums/Automation-questions/Rovo-Agent-Response-String-To-JSON/qaq-p/3045634

[Martin Fischer]

Hi @Jordi Romera,

I did a quick test with a simple Rovo Agent that just returns a Confluence documentation link, and another one that grabs any link it finds within the Jira work item. Both worked well in Jira comments — plain HTTP/HTTPS links are automatically detected and rendered as clickable, no extra formatting needed.

For example, I used a prompt like:

“You are an assistant that takes information from Confluence about the Azure Sync app, providing a link to the documentation, and returns the link in a format that can be clickable within Jira comments.”

or

“You are an assistant that takes any link you can find within this Jira work item and returns it in a format that is clickable within Jira comments.”

That was enough for Jira to handle the link properly:

Of course, if you need more control over formatting (e.g., forcing specific markup or richer output), you might want to look into creating a custom Rovo Agent using Forge. That gives you finer-grained control of the response payload.

[Jordi Romera]

Hello community! 👋

I'm having trouble getting a Rovo agent to properly add a comment with an embedded URL and transition a ticket from "Open" to "In Progress" through an automation rule.

What I'm trying to achieve:

1. Transition the issue to "IN PROGRESS" status
2. Add a comment with an embedded link in Markdown format: [Atlassian](https://www.atlassian.com/)

The problem: The automation rule runs, but either:

• The comment is not being added correctly
• The URL is not being embedded properly (appears as plain text instead of a clickable link)
• The transition is not working

What I've done so far:

• Set up an automation rule with "Use Rovo Agent" action
• Configured the agent prompt following the Markdown link format [text](url) as recommended
• Added the transition and comment instructions in the prompt

I'm attaching screenshots of:

• My automation rule configuration:

• The Rovo agent prompt I'm using

Could someone please review and tell me what I'm doing wrong? Any help would be greatly appreciated!

---

This is the Rovo agent prompt:

You are a Rovo automation agent for Jira Cloud. Execute the following tasks automatically and return a structured summary.

**Issue Context:**
- Issue Key: {{issue.key}}
- Summary: {{issue.summary}}
- Current Status: {{issue.status}}
- Assignee: {{issue.assignee}}

**Automated Tasks to Execute:**

1. **Transition to IN PROGRESS**
- Move issue from current status to "IN PROGRESS"
- Skip if already in this status

2. **Add Comment with Embedded Link**
- Add comment: "🤖 Automated update: Check out [Atlassian](https://www.atlassian.com/) for more information"
- IMPORTANT: Use Markdown link format: [text](url)
- Do NOT use Wiki Markup [text|url] or HTML <a> tags
- Ensure the link is properly embedded in the word "Vorwerk"

**Link Formatting Rules:**
- When referencing any URLs, always format them as clickable Markdown links
- Format: [Link Text](https://example.com)
- Never use raw URLs or HTML tags

**Output Format:**
Return your response as a structured summary that will be stored in {{agentResponse}}:

SUCCESS: [Action 1 result] | [Action 2 result]
or
ERROR: [Error description]

**Example Outputs:**
- "SUCCESS: Transitioned to IN PROGRESS | Comment with embedded Vorwerk link added successfully"
- "SUCCESS: Already in IN PROGRESS | Comment with embedded Vorwerk link added successfully"
- "ERROR: Transition failed - invalid workflow state"

**Important:**
- Execute all tasks automatically without asking for confirmation
- Always return a clear status message in plain text (no markdown in the response itself)
- The Markdown formatting should only be in the Jira comment, not in {{agentResponse}}

---

Thank you in advance!

[Martin Fischer]

Hello @Jordi Romera 

I’m not entirely sure what the benefit of using a Rovo agent is in this case — what part do you expect to be AI-driven? :) 

From what you’ve described, it seems like this could be handled entirely through automation. (Please correct me if I’m wrong!):

If you do want to stick with the Rovo solution, there are two possible issues to look into:

1. I’m not sure if you’ve set up Rovo Actions when building your agent — these define what the agent is actually allowed to do (like transitioning or commenting).
2. I’m also not sure if those actions can run inside automation without the confirmation step. In chat, you always have to confirm actions since AI can sometimes return unexpected results.

Workaround:
Let automation handle the Jira operations (transition, add comments)  .. 

but If you really want everything to be executed directly by the agent (within automation), you’d likely need to go the Forge route — where you can define custom Rovo actions to perform transitions or comments without the confirmation step. That approach can work, but it’s usually best reserved for controlled use cases.