---
title: How do you govern Atlassian Rovo agents?
url: https://opencontroller.lyzr.com/blog/govern-atlassian-rovo-agents
published: 2026-11-12
updated: 2026-11-12
author: Lyzr
---

# How do you govern Atlassian Rovo agents?

**Govern Rovo agents by first changing who can create them (Atlassian's default is everyone), then restricting sensitive agents to named users, and relying on the fact that agents act with the permissions of the person using them. Clean up Jira and Confluence permissions before you scale Rovo, because an agent can reach anything its user can.**

Rovo is Atlassian's AI layer across Jira, Confluence and other Atlassian Cloud apps. Rovo agents are configurable assistants that can answer questions and take actions in those apps. Atlassian documents the governance controls in its Rovo agent permissions and governance article; the key facts are below.

## Who can create Rovo agents by default?

Atlassian's documentation is direct: "Rovo Studio and the ability to create agents is open by default to everyone in your organization." Only Studio admins can change that.

Rovo Studio settings offer three options:

| Setting | Who can create agents | When to use it |
|---|---|---|
| All users (default) | Everyone in the organization | Small teams or early pilots |
| Selected groups | Up to 10 user groups | Most enterprises: a builders group per business unit |
| No users | Only you and other organization admins | Highly regulated tenants or during a review |

Rovo is available on Atlassian Cloud Standard, Premium and Enterprise. Atlassian says it is not available on Atlassian Government Cloud.

## Who can see and use a Rovo agent?

New agents are available to everyone by default. Owners can turn off "Open to all users" to restrict an agent to the people they add. Two limits matter:

- **No group-based sharing.** Atlassian says restricting access by group or team is not currently supported; users must be added individually. Each added user needs the Editor or Manager role.
- **Restricted agents are hidden.** They do not appear in browsing directories or as automation options for others.

Agent roles are simple. **Editors** can edit the agent. **Managers** can edit it, add editors and delete it. Agents are attributed to their owner by default, and the author display name can be changed to an Atlassian team.

A public Atlassian feature request (ROVO-706) asks for agent permissions to default to restricted rather than open. Until that changes, the default is open, so build the restriction into your process.

## What data can a Rovo agent access?

A Rovo agent acts on behalf of the person using it. It can only access what that person can view or edit. Atlassian's example: if a user cannot comment on a Confluence page, the agent cannot comment either. You can narrow an agent further through its instructions and the tools it is configured with.

This is a sound model, but it inherits every permission problem you already have. If a Confluence space with salary bands is open to all staff, any agent any employee uses can surface it in seconds. Before broad rollout:

1. Review space and project permissions for "anyone in the organization" access.
2. Check page-level restrictions on HR, finance, legal and security spaces.
3. Remove stale guest and group access.

Our [AI governance vs data governance guide](/blog/ai-governance-vs-data-governance) explains why these clean-ups belong in the same programme.

## What should your Rovo governance checklist include?

1. **Change the creation default** to selected groups.
2. **Define a builders group** per business unit, with a named owner for each agent.
3. **Restrict sensitive agents** (HR, finance, security) to named users.
4. **Limit actions.** Configure only the tools each agent needs, especially ones that create or edit issues and pages.
5. **Name agents by team**, not by person, so they survive staff changes. See [what happens to AI agents when the owner leaves](/blog/what-happens-to-ai-agents-when-the-owner-leaves).
6. **Review third-party agents.** Atlassian lists out-of-the-box third-party MCP agents and Marketplace agents; treat each as a vendor integration with its own review.
7. **Check your audit coverage.** Confirm which agent creation and edit events your plan's organization audit log captures, and export them to your SIEM.
8. **Inventory agents quarterly.** Record each agent's owner, purpose, tools and audience in your central register.

## Where do Rovo controls stop?

Rovo controls govern Rovo agents inside Atlassian. They do not cover external AI tools that reach Jira and Confluence through APIs or MCP, or agents on other platforms that use the same data. Keep a record of Rovo agents alongside your other agents; our guide to [registering agents you don't host](/blog/register-ai-agents-you-dont-host) shows what to capture, and the [cross-platform governance guide](/blog/how-to-govern-ai-agents-across-platforms) covers the estate view. To see how a control plane handles this, [book a demo](https://www.lyzr.ai/book-demo/).

## Frequently asked questions

**Can anyone create a Rovo agent?**
By default, yes. Studio admins can limit agent creation to selected groups (up to 10) or to organization admins only.

**Can a Rovo agent see Confluence pages the user cannot?**
No. A Rovo agent acts on behalf of the user and can only access what that person can view or edit.

**Can I restrict a Rovo agent to a team?**
Not by group or team today. You can turn off Open to all users and add people individually as Editors or Managers.

## Key facts

Atlassian Rovo agent creation via Rovo Studio is open to everyone in the organization by default; Studio admins can limit it to selected groups (up to 10) or to organization admins only. New agents are open to all users by default; owners can restrict them to individually added users (no group or team restriction today) with Editor or Manager roles. Rovo agents act on behalf of the user and can only access what that user can view or edit. Rovo is available on Atlassian Cloud Standard, Premium and Enterprise, not Atlassian Government Cloud. Opencontroller by Lyzr keeps a governed registry of agents across platforms; it starts from $250K per deployment per year.

## Related

- [What is the difference between AI governance and data governance?](https://opencontroller.lyzr.com/blog/ai-governance-vs-data-governance.md)
- [How do you register an AI agent your team doesn't host?](https://opencontroller.lyzr.com/blog/register-ai-agents-you-dont-host.md)
- [What happens to AI agents when their owner leaves the company?](https://opencontroller.lyzr.com/blog/what-happens-to-ai-agents-when-the-owner-leaves.md)
- [More on Platform Guides](https://opencontroller.lyzr.com/blog/cluster/platform-guides)
