Atlassian Rovo Pricing and Rovo Credits Cost with the Security Controls to Budget
Try it live
Watch AgentShield block an attack in real time.
Pick a scenario and drive the inspection lane yourself. No signup needed.
Run a request
Runs the live engine on your text. Nothing is stored, no account needed.
Inspection lane
INSPECTINGPolicy trace
High-risk action held for approval
Audit trail
- § · → → →
Atlassian Rovo has no separate license. It is included in paid Jira, Confluence, Loom and collection subscriptions through a monthly pool of Rovo credits, and the pool is shared across the whole organization. Per user per month, Jira and Confluence include 25 credits on Standard, 70 on Premium and 150 on Enterprise, and the Teamwork and Service collections include 250, 700 and 1,500. Extra usage is priced at 0.01 USD per credit, and billing for it starts on December 3, 2026.
Everything below was read on Atlassian's Rovo usage limits page on October 28, 2026. Third-party pricing guides disagree with each other and with Atlassian, so use these figures to question a quote, and confirm them on your order form.
Atlassian Rovo pricing at a glance
| Item | What Atlassian publishes |
|---|---|
| Standard, Jira or Confluence | 25 credits per user per month |
| Premium, Jira or Confluence | 70 credits per user per month |
| Enterprise, Jira or Confluence | 150 credits per user per month |
| Teamwork Collection or Service Collection | 250 on Standard, 700 on Premium, 1,500 on Enterprise |
| Pooling | Allowances from every app combine into one organization pool. Per-user allocation is listed as coming soon |
| Reset | Monthly on the billing date, unused credits do not roll over |
| Extra usage list price | 0.01 USD per credit, 10 USD per 1,000, with prepaid packs and annual volume discounts |
| Requirement | A verified business domain. Atlassian cannot enable Rovo on sites registered with a generic email domain |
The pooling rule changes the math. Atlassian's own example is 100 Confluence Premium users plus 100 Jira Premium users, which gives 14,000 credits a month in one pool. A few heavy users can drain it for everyone else.
What consumes Rovo credits
Search inside Atlassian apps, definitions and summaries are free. A chat message with a quick answer costs 10 credits, and a basic agent invocation costs 10 credits. Advanced agents, the ones where you pick a model, and features such as the Jira coding agent use variable credits that depend on reasoning steps, search volume, tools invoked and model power.
The MCP server is the line that matters most for security teams. Atlassian says installing the Rovo MCP server is free, and so are write operations and single-product lookups. Enriched calls, including the unified search tools, use variable credits at roughly one to ten per call. Read that carefully. The writes your AI clients make are the actions with the most security risk, and they are the ones Atlassian does not meter.
A worked example for a 200-person engineering organization
Take 200 people on Jira Premium and the same 200 on Confluence Premium. Using Atlassian's published numbers, that is 200 times 70 plus 200 times 70, which gives a pool of 28,000 credits a month. At 10 credits a chat message or basic agent run, the pool covers 2,800 such events. Spread across 200 people that is 14 events each per month, so a team that uses an assistant daily will pass it. Every 10,000 credits beyond the pool costs 100 USD at the list price. That is a small number for a company, and the reason to set a cap is the default behavior, not the amount.
The security conclusion is the opposite of the pricing one. The cheapest part of the AI bill is the part that grants an assistant write access to your tickets and pages. No credit meter will tell you when that access is misused, so the review has to come from controls on the action itself.
The December 3 extra usage change
Atlassian notifies admins at 80 and 100 percent of the allowance. Extra usage is on by default, which means features keep working past the allowance until a spending cap you set is reached. If extra usage is off, billable interactions pause at the limit. The current pricing table applies from August 31, 2026, and extra usage billing for Rovo credits takes effect on December 3, 2026. Set the cap before that date, because the default behavior is to keep spending. Teams that connect AI clients and agents to the MCP server should have procurement compare the order form with the cap you set before renewal.
The security lines outside the credits
| Line item | Included in Rovo credits | Where it comes from |
|---|---|---|
| Allowed domains, API token switch, IP allowlists | Yes, admin settings | Atlassian Administration |
| Data security policy blocking agents from covered content | No, needs Atlassian Guard Standard | Atlassian Guard. Classification coverage needs Guard Premium |
| MCP usage logs and audit logs | Yes | Atlassian |
| Review of every write an AI client makes on a signed-in user's behalf | No | A gateway rule layer such as AgentShield for Atlassian Rovo |
| Hold on comments to externally visible tickets | No | Argument rules and a named approver |
| One audit record across Atlassian and every other system your agents touch | No | A gateway audit trail |
Budget Guard if you want the read-side data security policy, because that control is the one Atlassian licenses separately. Budget the write-side review separately again, because no Atlassian line covers it.
Is the credit model a security problem
Only indirectly. Credits control spending, not behavior. An agent that creates two hundred issues through the MCP server draws no credits for the writes, so the bill will not warn you about it. The signal that something went wrong is in the usage logs, not the invoice. That is why a spend cap and an action-level rule do different jobs, and why neither replaces the other. Our human approval for AI agents guide explains how to choose which actions deserve a person.
When to buy nothing beyond the credits
| Your situation | What to do |
|---|---|
| Staff use Rovo search and summaries inside Atlassian apps | Buy nothing extra. Those features are listed as free |
| AI clients only read Jira and Confluence content | Use domain controls, turn API tokens off and consider Guard for the data security policy |
| AI clients create issues, edit pages or comment on tickets | Add an action-level control with approval on high-impact writes |
| Agents you build also write to billing, email or code systems | Add a gateway so one rule set and one record cover every system |
Two of the four rows need nothing from us, and that is the right answer for them.
What to ask Atlassian before you sign
- Which credit allowance applies to each of our apps, and what is the combined pool?
- What is our spending cap for extra usage, and who receives the 80 percent alert?
- Is the data security policy included in our Guard tier, and does it cover the classifications we use?
- Which MCP clients are allowed by default on a new site, and can we start with none?
- How long are MCP usage logs retained, and can we export them?
If your team will connect AI clients to Jira and Confluence, read our full Atlassian Rovo security page for the controls and the limits that API tokens leave open.
See the firewall block an attack live.
Drive the Threat Console and watch a real prompt injection get stopped, then put AgentShield in front of your own agents.