What is the Atlassian Rovo MCP Server?
The Atlassian Rovo MCP (Model‑Connect‑Publish) server is a Docker‑packaged service that exposes secure APIs for Atlassian products such as Jira and Confluence. It enables AI assistants and autonomous agents to create, read, update, and delete work items directly from those tools.
- Create and update Jira issues, epics, and stories
- Generate, edit, and publish Confluence pages
- Expose standard MCP endpoints that any MCP‑compatible client can consume
How to enable the Atlassian Rovo MCP Server with Docker
Docker’s MCP Catalog and Toolkit provide a one‑click workflow that eliminates manual installation and dependency management.
- Prerequisites
- Machine with at least 8 GB RAM (16 GB recommended)
- Docker Desktop installed and running
- Open Docker Desktop → MCP Toolkit tab
- Search the Docker MCP Catalog for “Atlassian Rovo MCP Server”
- Select the remote version (cloud icon) and click **Enable**
- Docker automatically pulls the container image, configures OAuth credentials, and starts the service
After the server is running, connect any MCP‑compatible client (Claude Desktop, Gemini CLI, Codex, etc.) by clicking **Connect** in the client UI or by using the provided CLI command.
Why use the Atlassian Rovo MCP Server with Docker
Running the server in Docker delivers three core advantages:
- Reduced operational friction – No manual dependency resolution, environment variables, or network configuration.
- Secure, built‑in authentication – Docker’s OAuth integration stores credentials safely and reuses them across all MCP clients.
- Client‑agnostic connectivity – The same server instance can be used by any AI agent without re‑configuration, enabling seamless switching between tools.
Seamless authentication with built‑in OAuth
The server relies on Docker’s OAuth provider. Users authenticate once through Docker Desktop; the token is then injected into the MCP container, allowing all connected agents to act on behalf of the same Atlassian account.
Working with any MCP‑compatible AI agent
Because the server follows the standard MCP contract, integration steps are identical for all agents:
- Open the agent’s settings panel
- Choose “Connect to MCP server”
- Select the Atlassian Rovo instance from the dropdown
- Confirm and restart the agent if required
After connection, agents can issue commands such as “Create a Jira epic for the new feature” or “Update the Confluence roadmap page”.
Frequently Asked Questions
- Is the server local‑only or remote? Both options exist; the quick‑start described here uses the remote version hosted in Docker’s MCP Catalog.
- Do I need to manage credentials? No. Docker’s OAuth layer handles token storage and rotation automatically.
- Can I share a configured workflow with teammates? Future releases will allow exporting a complete MCP configuration as a shareable package.