Rovo chat agent's response

Question

Hi Team,

Please check agent's response where it is either sharing URL or sharing different steps. It should share the exact steps from the document

Why is it so?

{
"author": {
"id": "643160dc-0f1f-44a7-9c3b-363ce214f48d",
"actor_type": "ASSISTANT",
"ari": "ari:cloud:identity::user/712020:0c8c6855-612f-47c3-b607-2c219ecf6f20",
"named_id": null,
"name": "DevOps Genie"
},
"scenario": {
"id": "89cd6ac9-399b-428f-b616-5054b97b0b7d",
"name": "Local Issue",
"actionIds": []
},
"message_metadata": {
"requestId": "271db7b0-1aa6-413c-81b7-4a23c380a911",
"workflowId": "labs"
},
"actions": [],
"plugin_invocations": []
}

Dr Valeri Colon (Connect Centric) · Answer

Hi @Nachiyar.M in response to your initial question: Yes—this behavior is expected unless the agent is explicitly constrained. Rovo agents don’t automatically quote documents step-by-step; they summarize or rephrase based on understanding. If the instructions allow flexibility, the agent may share links or inferred steps instead of exact text. To force exact steps, your prompt and scenario must clearly say things like “respond only using the provided document” or “do not paraphrase.” Without that, Rovo optimizes for helpfulness, not verbatim accuracy.

To answer your follow-up questions:

No—as the agent creator/admin, you generally can’t view other users’ full chat histories for privacy/security. You may have limited usage/analytics depending on your plan and org settings, but not verbatim conversations.

Atlassian hasn’t published a hard public limit for number of scenarios. Practical limits may apply (UI/manageability or org policies), so treat it as “supported, but keep it simple.”

Agent instructions, scenarios, and knowledge are stored in Atlassian Cloud. Conversation history is tied to the user/session (and org retention settings). Heavy usage shouldn’t slow the agent itself, but very long threads can cause timeouts—start new chats for best results.

Nachiyar.M · Answer

Hi Team,

Thank you for responding. The previous issues were resolved by changing the prompts & scenario instructions.

I have few queries, please check if you can help me.

1. I have created a Rovo agent & given a custom knowledge & training it. While I ask my colleagues to test the agent's response they have their conversation history. But is there any way I can read the logs of the other user who are using my Agent (Me being the admin/creator)

2. Do we have any restriction in the number of scenarios we create?

3.  All these instructions/scenarios/conversation history - Where is it getting stored? If huge volume of users use the agent, will there be any performance issue?

Not sure where to post my queries, please let me know if you can guide me the ways to reach for this kind of support.

Thanks :)

Nachiyar.M · Answer

Can you please share the link if you have any to contact Atlassian support?