MCP Atlassian is a Model Context Protocol server that connects AI assistants to Atlassian’s Confluence and Jira. With it, an MCP client such as Claude can search issues, read documentation pages and create or update work items using your own Atlassian account. The project has about 6,000 stars on GitHub, is MIT licensed, and was actively developed in 2026.
The server exposes Jira and Confluence as tools an LLM can call: JQL searches, issue creation and transitions, page retrieval and authoring, and more. It supports both Atlassian Cloud and self-hosted Server/Data Center deployments, and authenticates with API tokens, personal access tokens or OAuth depending on your setup. Safety controls include a read-only mode that blocks write operations entirely. Installation is a one-liner via the published PyPI package (uvx mcp-atlassian) or Docker, so it drops into existing MCP client configurations in minutes.
In practice, the server turns an AI assistant into a junior teammate with supervised access to your project tooling: it can triage a Jira board, pull the relevant Confluence specification into context before writing code, or draft an issue from a conversation. Administrators keep control through scoped tokens and the read-only switch, which makes it one of the safer MCP servers to pilot inside a company. It is community-maintained rather than an official Atlassian product, though it is the most widely adopted of the Atlassian MCP servers.
Free. MCP Atlassian is open source under the MIT license; you use it with your existing Atlassian subscription and credentials.
Teams living in Jira and Confluence who want their AI assistant to search tickets, summarize documentation and update work items without manual copy-pasting.