Atlassian MCP Server

What atlassian-mcp-server MCP server does

The atlassian-mcp-server MCP server gives compatible AI applications access to Atlassian Cloud data and actions through the Model Context Protocol. It is an official, cloud-hosted bridge rather than a local connector that runs against an on-premises Atlassian installation.

Supported Atlassian products and data sources include Jira, Confluence, Jira Service Management, Bitbucket, Compass, Loom, and platform data such as Projects, Goals, Teams, Focus, Talent, and the Teamwork Graph. The material describes use cases including searching and summarizing records, reviewing Loom recordings, and creating or updating work items and pages from natural-language instructions.

How it works

The client connects to Atlassian's remote MCP endpoint and presents the server's available actions to an AI model. The model can then select actions based on a user's request, such as finding project information, reviewing documentation, or turning meeting notes into work items.

The recommended v2 endpoint is:

https://mcp.atlassian.com/v2/mcp

Authentication is handled with OAuth 2.1 or Atlassian API tokens. Requests operate with the authenticated user's existing access controls, so the server does not describe a separate permission model that bypasses Atlassian access restrictions.

Setup and configuration

Use an MCP client that supports remote HTTP MCP connections, then add the v2 endpoint or use the client's native installation flow. The repository provides one-click setup links for Cursor and VS Code, and describes connections for Claude, ChatGPT, Cursor, VS Code, and other AI tools.

New configurations should use the v2 URL because it exposes more tools and products than the earlier version. Existing v1 connections are moved toward v2 tools automatically. The README notes that some clients may need cached client IDs or .well-known credentials cleared if authentication fails after this transition.

No local package, runtime command, or environment variable is specified in the provided material. The server is hosted by Atlassian, so setup consists of configuring the remote endpoint and completing the supported authentication flow.

Tools and capabilities

The atlassian-mcp-server MCP server supports capabilities described at a product level rather than as a complete tool manifest in the supplied material:

• Search and summarize Jira, Jira Service Management, Confluence, Bitbucket, Projects, Goals, and related Atlassian data.
• Retrieve and review Loom video and meeting recordings.
• Create and update Jira work items or Confluence pages.
• Automate repetitive workflows, such as generating work items from meeting notes or specifications.
• Access Compass and additional Atlassian platform data through the Teamwork Graph.

The exact set of exposed tools can vary by server version; the repository identifies v2 as the recommended release.

Limitations and notes

This integration is intended for Atlassian Cloud. The provided material does not document self-hosted deployment, local execution, or support for non-Atlassian systems. Results and mutations are subject to the permissions of the authenticated Atlassian user.

Although the repository is open source, the MCP service itself is described as an Atlassian-hosted cloud service. Client compatibility matters during the v1-to-v2 transition, particularly for clients that cache authentication metadata. The README also points to separate Atlassian documentation for the current supported-tool list and security administration details.