atlassian/atlassian-mcp-server

Official remote MCP server for Atlassian. Securely connect Jira, Confluence, Jira Service Management, Bitbucket, and Compass to Claude, ChatGPT, Cursor, VS Code, and other AI tools using OAuth 2.1 or API tokens.

Atlassian Rovo MCP Server – what it is

The Atlassian Rovo Model Context Protocol (MCP) server is a cloud‑hosted bridge that lets AI‑enabled tools (ChatGPT, Claude, VS Code Copilot, Cursor, etc.) talk directly to the data in an Atlassian Cloud site – Jira, Confluence, Bitbucket, Compass, Loom and the broader “Teamwork Graph” (projects, goals, teams, talent, …). It implements the open Model Context Protocol so any MCP‑compatible client can discover and invoke Atlassian‑specific actions without the user having to copy‑paste URLs or switch tabs.

Key capabilities

Capability	What it lets you do
Real‑time read/write	Search, retrieve, and update issues, pages, repos, videos, etc., from within an AI chat or coding assistant.
Natural‑language commands	Describe an action (e.g., “Create a Jira ticket for the bug in the latest PR”) and the server translates it into the appropriate Atlassian API call.
Secure authentication	Uses OAuth 2.1 or API‑token auth, so every request respects the user’s existing Atlassian permissions.
Fine‑grained scopes	Permission groups such as `read_jira`, `write_confluence`, `delete_jira`, etc., are exposed as separate tools; admins can enable/disable them per‑product.
Tool discovery	Only a small set of “primary” tools are sent up‑front; the client can request additional tools on demand, keeping the AI’s context lightweight.
Version 2 endpoint	The current recommended URL is `https://mcp.atlassian.com/v2/mcp`, which adds more products and tools over the older v1 endpoint.

Who would use it?

Developers who want their IDE or Copilot‑style assistant to create or update Jira tickets, pull‑request comments, Confluence pages, etc., without leaving the editor.
Content creators / product managers who want to ask ChatGPT to summarise a Confluence space or pull a Loom video transcript.
Automation engineers building custom agents that need to read/write Atlassian data as part of a larger workflow.

How it works (high‑level flow)

Install the connector – One‑click buttons add the Atlassian MCP plugin to the chosen AI client (Cursor, VS Code, ChatGPT, Claude, …). The plugin contains a tiny manifest (plugin.json, .claude-plugin/plugin.json, etc.) that points to the server URL.
Authenticate – The user runs the OAuth 2.1 flow (or supplies an API token). The server receives an access token that encodes the user’s Atlassian permissions.
Tool discovery – The client asks the server for the list of available tools. Primary tools are returned immediately; others can be fetched later via the MCP discover call.
Invocation – When the AI decides a tool is needed, it sends a request to the server (e.g., createIssue, searchConfluence). The server translates the request into the appropriate Atlassian REST API call, forwards it, and returns the result.
Result handling – The AI incorporates the response (search results, created IDs, video transcripts) into its reply to the user.

Supported products & permission groups

Product	Read	Write	Search	Delete / Manage
Jira	✅	✅	✅	`delete_jira`, `manage_jira` (admin‑enabled)
Confluence	✅	✅	✅	—
Bitbucket Cloud	✅	✅	—	—
Jira Service Management	✅	✅	—	(API‑token only)
Compass	✅	✅	—	—
Loom	✅	✅	—	—
Teamwork Graph (Projects, Goals, Teams, Focus, Talent)	✅	✅	✅	—

Each group maps to an OAuth scope; admins can enable or disable them per‑organization.

Getting started (quick checklist)

Have an Atlassian Cloud site with the products you need.
Pick an MCP‑compatible client (ChatGPT, Claude, VS Code, Cursor, etc.).
Click the Add to … button in the README table – the client will install the Atlassian plugin and pre‑populate the server URL (https://mcp.atlassian.com/v2/mcp).
Follow the OAuth flow (or paste an API token) to grant the AI access.
Start using natural‑language commands; the AI will automatically discover and call the appropriate Atlassian tools.

Where to find more info

Full setup guide – https://support.atlassian.com/atlassian-rovo-mcp-server/docs/getting-started-with-the-atlassian-remote-mcp-server/
Supported tools reference – https://support.atlassian.com/atlassian-rovo-mcp-server/docs/supported-tools/
Security & admin details – https://support.atlassian.com/security-and-access-policies/docs/understand-atlassian-rovo-mcp-server/
Community & support – Atlassian Community forums linked in the README.

TL;DR

The Atlassian Rovo MCP Server is a production‑grade, cloud‑hosted service that implements the open Model Context Protocol, allowing AI assistants to read and write Atlassian Cloud data securely and in real time. Install the appropriate connector, authenticate, and then let your AI agent issue natural‑language commands that are turned into Jira, Confluence, Bitbucket, Loom, Compass, etc., operations.