The Atlassian Rovo MCP Server is a cloud-hosted Model Context Protocol (MCP) server that connects AI-powered clients to your Atlassian Cloud apps. It provides a standardized interface for AI assistants, IDEs, and automation tools to read, create, and update work across Jira, Confluence, Jira Service Management, Bitbucket, Loom, and the wider Atlassian platform - all secured by OAuth 2.1 authorization and your existing Atlassian permissions.

What you can do

With the Atlassian Rovo MCP Server, AI clients can:

Search and summarize Jira work items, Confluence pages, Bitbucket pull requests, and other Atlassian content without context switching.
Retrieve Loom recordings, including transcripts, comments, AI briefs, and extracted meeting action items.
Create and update work items, pages, and other content using natural language commands.
Automate repetitive work, such as generating Jira work items from meeting notes or creating Confluence documentation from specs.
Query across apps using Teamwork Graph, such as finding the pull requests, deployments, and documents connected to a work item.

How it works

The MCP Server acts as a bridge between MCP-compatible clients and Atlassian Cloud APIs:

An AI client connects to the server endpoint at https://mcp.atlassian.com/v2/mcp.
The server initiates a secure OAuth 2.1 authorization flow (or accepts an API token for headless authentication).
Once authorized, the client can invoke tools to interact with Atlassian apps.
All actions respect the authenticated user's existing permissions - the server never grants access beyond what the user already has.

Rather than sending every tool definition at connection time, the server advertises a small set of primary tools and lets clients discover the rest of the catalog on demand through the discover tool. This preserves context window and lets Atlassian add tools without requiring you to reconnect. See Supported tools for details.

Supported apps

App	Capabilities
Jira	Search, create, edit, transition, and comment on work items; query spaces, boards, sprints, versions, and metadata
Confluence	Read, create, and update pages, live docs, whiteboards, and databases; search with CQL; manage comments, attachments, permissions, and spaces
Jira Service Management	Query and update ops alerts, on-call schedules, and team information
Bitbucket Cloud	Browse and act on repositories, pull requests, pipelines, deployments, and environments
Loom	Read and manage videos, transcripts, comments, meeting action items, and folders
Goals, Projects, Focus areas, Teams, Talent	Search, read, create, and update Atlassian platform work-tracking entities
Teamwork Graph	Retrieve connected context and relationships for any Atlassian entity
Rovo Search	Semantic search across Jira, Confluence, and connected apps using natural language
Code search	Search and read source code across connected code providers

Rovo credit usage

Some tools consume Rovo credits. Consumption depends on:

Context volume - retrieving larger amounts of data costs more credits.
Reasoning depth - complex queries consume more credits than simple lookups.
Usage caps - per-call limits help manage cost. For example, Teamwork Graph and Rovo Search tools consume up to 10 credits per call.

Supported clients

The server works with any MCP-compatible client, including:

OpenAI ChatGPT
Claude Desktop and Claude Code
Docker
GitHub Copilot CLI
Google Gemini
Amazon Quick Suite
IDEs such as VS Code, Cursor, and Windsurf