Is it possible to set Rovo agent access permissions by group?

[田中優貴]

Is it possible to set Rovo agent access permissions by group?

Body
I’m currently creating multiple agents in Atlassian Rovo Studio.
My goal is:

• Allow everyone in the organization to use Rovo
• But restrict access to specific agents by group
  • Example: Sales Department agent → only members of the Sales group can use it
  • IT Department agent → only members of the IT group can use it

Right now, in the “Users and permissions” section when creating an agent, I can add individual users, but I don’t see an option to add groups.

I’d like to confirm:

1. Is there any way to set agent access permissions by group?
2. If not, are there any workarounds (e.g., API or CLI to bulk add group members)?

Environment:

• Atlassian Cloud
• Rovo Studio enabled
• Admin permissions available

If anyone has faced the same challenge or knows the official guidance, please share.

[Jensen Fleming · Accepted solution]

Hey @田中優貴 -- The "users and permissions" that currently exist on Agents, is for "editing/managing" the agent, not for "using" the agent. 

Currently, anyone with access to Rovo/AI in your site, can use any agent. There is no blocker here. 

We will be adding "who can use this agent" permission in the near future. It will likely not support Groups/Teams yet, but we will most likely be adding a CSV uploading feature to let you do bulk.  You can follow along for that feature here: https://jira.atlassian.com/browse/ROVO-142

[田中優貴]

Thanks for your answer.

[David Ingram]

Same problem, here is my explanation (FEATURE REQUEST):

Feature Request: Space-Specific Agent Visibility and Management

Summary

Provide administrators with the ability to control which AI agents are available within each Confluence Space, including:

• Restricting agent visibility to specific Spaces (or Selecting which agents are displayed within a Space) (or Hiding all unwanted agents from users within a Space)
• Defining the display order of agents within a Space
• Setting a default agent for a Space

Problem Statement

Today, when a custom agent is created for use within a specific Confluence Space, that agent becomes available alongside all other agents that have been created across the organisation in all Spaces

This creates several challenges:

1. Users can easily select the wrong agent.
2. Space-specific agents are diluted amongst unrelated agents.
3. Users receive inconsistent or inappropriate results because the selected agent may not contain the correct instructions, tools, skills, knowledge sources, or conversation starters.
4. Administrators have no way to guide users towards the intended experience.
5. As the number of agents grows, discoverability and governance become increasingly difficult.

The issue already exists today and will become significantly more problematic as organisations create larger numbers of specialised agents.

Proposed Solution

Introduce Space-level agent management controls that allow Space administrators to:

Agent Visibility

• Specify which agents are available within a particular Space.
• Restrict agents to be available to show/not-show in one or more designated Spaces (or Hide agents that are not relevant to the Space).

Agent Ordering

• Control the order in which agents are displayed to users.
• Promote preferred or recommended agents to the top of the list.

Default Agent Selection

• Designate a default agent for each Space.
• Automatically present the most relevant agent when users enter the Space.
• End user can override.
• Remember last Agent selection for returning user.

Governance Controls (this feature enables)

• Preventing users from accessing agents that have not been approved for that Space.
• Allowing organisations to create curated AI experiences tailored to different teams, departments, or business functions.

Business Impact

Many organisations will create specialised agents for different domains, teams, and use cases. Examples include:

• Product Management agents
• Engineering agents
• HR agents
• Sales agents
• Customer Support agents
• Project-specific agents

Each of these agents may contain different instructions, tools, knowledge sources, permissions, and expected behaviours.

Without Space-level control, users can inadvertently use (in fact are LIKELY to use) the wrong agent and receive responses that are incomplete, misleading, or inappropriate for their context.

Providing Space-specific agent management would:

• Improve user experience
• Increase trust in AI-generated responses
• Reduce user confusion
• Improve governance and compliance
• Enable more effective knowledge management
• Support large-scale organisational adoption of AI agents

User Story

As a Confluence Space Administrator,

I want to control which AI agents are visible, available, and prioritised within my Space,

So that users interact with the most relevant agent for their context and are not distracted or confused by unrelated agents from elsewhere in the organisation.

[Bob Dalm]

Hi @田中優貴, 

Currently this is not possible. From what I understood, this is planned for the foreseeable future though!

Best, 

Bob

[田中優貴]

Thank you so much for your response.