Community resources
Community resources
Community resources
Use Teamwork Graph in Rovo MCP Server (Open Beta)
TLDR: Two new tools in the Rovo MCP server – getTeamworkGraphContext and getTeamworkGraphObject – give any MCP-compatible AI agent access to Atlassian's Teamwork Graph: 80B+ relationships across Jira, Confluence, Bitbucket, Loom, JSM, and 75+ third-party tools. Available now in Open Beta: get connected and share feedback.
Working in the terminal? Teamwork Graph CLI gives coding agents like Claude Code and Cursor the same connected context, right where they build. Learn more here.
Hi Atlassian Community,
I’m Kalvin Xu, Teamwork Graph Product Manager, and today we’re announcing Teamwork Graph tools in Rovo MCP server!
We've brought Teamwork Graph tools to the Rovo MCP server, so your AI agents can see the full context across your work. That’s Atlassian data and 75+ third-party tools, all connected in one graph. And you don't have to be in Rovo to get it. Any MCP-compatible AI client works. Wherever you work, the context comes with you.
Why use Teamwork Graph tools in Rovo MCP?
Every team builds up a wealth of knowledge as they work – decisions captured in Confluence, progress tracked in Jira, code reviewed in Bitbucket, designs iterated in Figma, conversations held in Slack. The problem is that this knowledge is scattered across tools, buried in links, and locked behind context that only the people who were there truly understand.
Most MCP tools try to solve this by giving AI a way to fetch data - pull a ticket, retrieve a page, grab a list. But fetching data isn't the same as understanding it. And stitching isn’t the same as connecting.
A raw Jira issue tells you what the ticket says. It doesn't tell you what’s blocking it, who owns the related PR, what the Confluence spec says, or which Slack thread captured the decision that changed everything. It doesn't show you the people behind the work - who made the call, who's accountable, who's been silently unblocking it for weeks.
That's the difference between a lookup and a graph. Teamwork Graph connects work, knowledge, and people across Atlassian and 75+ third-party tools - in one unified graph, not a patchwork of point tools. With Teamwork Graph, you’ve got AI that can understand your team's context the way you do.
What's launching
Two new tools are now available in the Atlassian Rovo MCP server, usable by any AI agent that supports the Model Context Protocol - including Cursor, VS Code, Replit, and more:
-
getTeamworkGraphContext - give it a Jira issue, a Confluence page, a team member, or any other work, and it returns everything connected to that object in the Teamwork Graph: linked work items, related pages, pull requests, deployments, documents, collaborators, and more.
-
getTeamworkGraphObject - takes the connections discovered above and retrieves their full content: descriptions, statuses, comments, page content, timestamps, and authors.
The two tools work as a pair. getTeamworkGraphContext maps how work is connected. getTeamworkGraphObject reads the detail. Your agent chains them together - and can call getTeamworkGraphContext recursively to follow connections multiple hops deep, uncovering context that would take a person minutes of manual clicking through tabs and tools.
See it in action
Try it out
Start building with the full picture
"Summarize this project. What are the dependencies, what needs my action, and what needs to get done?"
Point your agent at a project ticket. Instead of just reading the ticket, it traverses the graph to find linked documents, related designs, open blockers, and where decisions stand today. The summary it produces is grounded in the actual state of the work — not a generic recap you'd have to fact-check yourself.
Design informed by what the team already knows
"I’m new to this project. Take a look at this ticket and design a v1 concept prototype for me."
A designer starting work on a new feature gives their agent a Jira epic. The agent discovers linked specs, competitive research buried in Confluence, technical constraints in architecture documents, and even a previous attempt at the same feature that was rolled back. The first design iteration is informed by months of team knowledge - assembled in seconds.
Prepare for meetings without the manual data gathering
"Help me prepare for my strategy session at 2PM. Tell me everything about project blueberry."
Before a planning session or a customer call, ask your agent about a project or account. It maps the connected work - delivery status across epics, who's been most involved, open risks, recent updates - across Jira, Confluence, and projects. You walk into the meeting with the complete picture, without having Slacked five people to piece it together.
Surface third-party context alongside Atlassian data
The Teamwork Graph doesn't stop at Atlassian products. When your team links Figma files, Google Docs, Slack threads, or GitHub repos to their Jira issues and Confluence pages, those connections appear in Teamwork Graph too. Your agent can surface a customer's requirements doc linked to an epic, or an archived design file linked to a closed issue - context that spans your entire toolchain.
Getting started
-
Set up the Atlassian Rovo MCP server in your AI agent environment. Follow the setup guide for step-by-step instructions.
-
Authenticate with your Atlassian account. The MCP server respects your existing permissions - your agent can only access data you already have access to.
-
Start asking questions. Reference a Jira issue, a Confluence page, or a team member. Your agent will use the Teamwork Graph tools to discover connections and pull in context automatically. You don't need to tell it which tools to use - just ask what you need to know.
For the full list of supported tools and detailed parameters, see the supported tools documentation.
What's ahead
These tools are available now in open beta, with support for Jira issues, Confluence pages, and Atlassian users as starting points. We're actively expanding the set of supported entity types and relationship categories - with broader object support.
Which AI client are you planning to use these with first? I'd love to hear how it changes your workflow. Let me know in the comments below
There’s more on the way - we're putting together a Bite-sized Learning to help you get hands-on fast. Watch this space!
FAQ
What data does the Rovo MCP server access and who can see it?
Teamwork Graph tools in Rovo MCP respect your existing Atlassian permissions. Your AI client only surfaces data that the authenticated user is already authorized to see. No new access is granted by connecting to the MCP server.
Is my data sent to third-party AI providers when I use these tools?
When you use an MCP-compatible AI client (like Claude or ChatGPT), data retrieved from Atlassian is passed to that client to generate a response. This is governed by your agreement with that AI provider - not Atlassian. We recommend reviewing your AI client's data handling policies before connecting sensitive workspaces.
Can admins control which users can connect to the Rovo MCP server?
Yes, admins can enable or disable the Rovo MCP server in Atlassian admin settings and control access at the org level.
Is the Rovo MCP server secure for enterprise use?
Yes, the Rovo MCP server is admin-controlled and uses OAuth for authentication - the same secure, governed auth flow used across Atlassian products. All data access is scoped to the authenticated user's existing permissions.
What authentication method does Rovo MCP use?
OAuth is supported today, along with API tokens. OAuth is the recommended method for enterprise deployments.
How much will it cost to use Teamwork Graph tools in Rovo MCP in the future?
Today, Teamwork Graph tools are available for free as part of the Rovo MCP server and are included with your Rovo entitlement.
Teamwork Graph MCP tools aggregate data across multiple Atlassian products and connected apps, and may use AI processing to generate contextual insights. When these tools move to General Availability, they will be billed at a minimum of 1 Rovo credit per call. Calls that involve AI inferencing or multi-step graph queries may be billed at a higher rate. We will provide customers at least 90 days' notice before any charges take effect, along with published pricing details.
What AI clients work with Teamwork Graph tools in Rovo MCP?
Any MCP-compatible AI client - including Claude (Anthropic), ChatGPT (OpenAI), Cursor, VS Code with GitHub Copilot, and more. If your AI client supports MCP, it can connect to the Rovo MCP server. For the official list of clients, go here.
What's the difference between getTeamworkGraphContext and getTeamworkGraphObject?
They work as a pair. getTeamworkGraphContext maps everything connected to a work item - linked issues, pages, PRs, collaborators, and more. getTeamworkGraphObject then retrieves the full content for those connected objects. Think of it as: discover first, then fetch.
What Atlassian data can these tools access?
Jira issues, Confluence pages, Atlassian goals, projects, JSM incidents, Looms, and more - plus connected third-party data like GitHub, Figma, Miro, and Slack (where connected).
What's the difference between Teamwork Graph tools in MCP and the Teamwork Graph CLI?
Complementary, not competing. MCP is best for web-based AI clients and IDEs (Claude.ai, ChatGPT, VS Code). CLI is best for agent terminals and coding agents (Claude Code, Cursor, Codex). Many teams will use both - pick the surface that fits where you work.
How is using Teamwork Graph CLI different from just searching Confluence or Jira?
Search finds documents. The Teamwork Graph finds connections. An agent using Teamwork Graph can start from a Jira issue and traverse to the architecture doc, the related PR, the team member who last updated the spec, and the design file in Figma - in a single query chain.
Was this helpful?
Thanks!
Kalvin Xu
2 comments