Clarification on Atlassian Rovo MCP Server Rate Limits (10,000 calls on Enterprise Plan)

[Kanwar Robinson]

Hi Team,

I was reading the Rovo MCP Server documentation, which states that Enterprise customers can make 1,000 calls per hour plus an additional 20 calls per user, up to a maximum of 10,000 calls per hour.

I have a question about how these calls are counted.

Suppose my MCP client sends one request to the Rovo MCP server, but internally the server performs multiple API calls (for example, 2–3 Jira or Confluence API requests) to fulfill that single client request.

In this scenario:

Is only the single client request counted against the 10,000-call limit?
Or are all internal API calls made by the Rovo MCP server also counted toward the rate limit?

[Viswanathan Ramachandran · Accepted solution]

Hi @Kanwar Robinson 

Atlassian’s docs describe the rate limit as calls to the Rovo MCP server, not the number of downstream Jira/Confluence REST requests the server may make internally to fulfil one tool invocation. So in practice, one client request to the MCP server should count as one MCP call against the 10,000/hour cap, even if the server fans out into multiple internal API requests behind the scenes. See Rovo usage limits 

That said, I’d treat this as the most likely interpretation rather than a guaranteed policy statement. The safest assumption is still that the MCP request is what’s metered, while the server’s own internal work may be separately constrained by its backend/service limits. 

You can see this -> Monitor Rovo MCP server activity  and hope it helps. 

[Rebekka Heilmann _viadee_]

I @Kanwar Robinson - welcome to the Community,

don't know for sure, but from the documentation and how Atlassian works elsewhere, MCP rate limits are separate from API rate limits. So internal API calls are not counted when invoked by the official Atlassian MCP.