MCP or Rest API for ROVO

[Aman Mehra]

Hi , i wanted to check if there any update on development of programatic access for Rovo ?

[Alexander Nilsson]

Hello Aman, 

thanks for your question!

Right now, there isn’t a public REST API that exposes Rovo itself (for example, a /rovo/chat endpoint you can call from any external service).

For programmatic scenarios, there are two main options today:

• For external AI tools and IDEs, Atlassian provides the Rovo MCP Server, which lets compatible clients securely access Jira, Confluence, and Compass data via the Model Context Protocol while respecting existing permissions.
• For in‑product automation and custom agents, you can use Forge with Rovo Agents/Actions and the Rovo bridge API (currently in early access), and still rely on the standard Jira/Confluence REST APIs for the underlying data.
• Documentation:
  • https://support.atlassian.com/atlassian-rovo-mcp-server/docs/getting-started-with-the-atlassian-remote-mcp-server/
  • https://developer.atlassian.com/platform/forge/apis-reference/ui-api-bridge/rovo/

If you can share a bit more about your use case, the community can point you to more concrete examples or patterns.

Greetings,

Alex

[Aman Mehra]

Thank you Alex! Much appreciated. 

The use case : search and summarize that is part of a custom built agent workflow and eventually can perform actions in Jira and Confluence. 

I did look at the docs, I am assuming integration with custom built agents is permissible apart from the list of tools already mentioned ?

Your response would be helpful, thanks again. 

[Alexander Nilsson]

Hi Aman,

sure, here is a concrete example flow and how you could implement it technically.

Example flow:

1. The user asks the agent: “Summarize all open bugs for Project XY and create a Jira summary ticket.”
2. The agent searches Jira for all open issues with issuetype = Bug in the “XY” project and groups them by component and severity.
3. The agent generates a short natural-language summary (key themes, most affected components, top critical bugs).
4. The agent creates a new Jira issue (for example type = “Task”) that:
   • Contains the generated summary in the description.
   • Links the identified bug issues as “relates to” or “is caused by”.
5. The agent replies in chat with the summary and a link to the new Jira ticket.

How this maps to a technical implementation:

• Define a Rovo Agent that understands prompts like “summarize open bugs for project XY” and routes to one or more actions.
• One action calls the Jira REST API (/rest/api/3/search) to fetch all open bugs for the given project and aggregates them by component and severity: https://developer.atlassian.com/platform/forge/build-a-jira-issue-analyst-rovo-agent/#step-4--update-index-jsx 
• The agent uses that aggregated data as context to generate the natural-language summary.
• Another action calls the Jira REST API (/rest/api/3/issue) to create the “summary” issue and then adds links back to the individual bugs.
• The agent then returns a chat response with both the readable summary and the link to the created issue.

For reference, the following docs walk through similar patterns step by step:

• Build a Jira issue analyst Rovo Agent – end-to-end Rovo Agent + Jira example 
• Build a Q&A Rovo Agent for Confluence – similar idea, focused on Confluence content
• Rovo bridge API reference – how the agent talks to your actions

You can adapt the above flow 1:1 for Confluence by swapping the Jira REST calls for the Confluence REST API (search pages → summarize → create a “status” page that links all sources).

Greetings,
Alex