Issue Summary

When using a Rovo Agent as a component in a Jira Automation rule, the execution intermittently fails or returns a blank response. This occurs when the agent decides to call a tool or action instead of providing a text-based answer. Since Automation currently runs agents without a "human-in-the-loop" and does not support auto-execution of agent actions, the run terminates without a final text response, leaving the {{agentResponse}} smart value empty.

Steps to Reproduce

1. Create a Rovo Agent with instructions that might lead it to use a tool (e.g., "Review this ticket and update the priority if it's urgent").
2. Set up a Jira Automation rule that invokes this Rovo Agent.
3. Use the {{agentResponse}} smart value in a subsequent action (e.g., Add Comment).
4. Trigger the rule. The failure is intermittent, occurring only when the LLM chooses a tool call over a text response.

Expected Results

The Rovo Agent should either auto-execute the action (if permitted) or always provide a text-based summary of its intent so that the automation flow can complete with a valid response.

Actual Results

The agent returns no text response because it is "stuck" waiting for a tool call confirmation that cannot be provided in an unattended automation context. The automation audit log may show "SOME ERRORS" or post a blank comment.

Workaround

• {}Modify the agent's instructions to explicitly state: "Do not use any tools or actions. Provide your response as plain text only."

• Automation Logic: Add a "Compare condition" before the comment action to verify that {{agentResponse}} is not empty.

• Logging: Add a Log action for {{agentResponse}} immediately after the agent step to verify if the response was empty during a failure.