An MCP server that turns LLMs into Jira power users

What it does

Wraps Atlassian’s APIs into 72 MCP tools, letting compatible AI assistants query JQL, read Confluence pages, create tickets, transition issues, and post updates. Works against both Atlassian Cloud and self-hosted Server/Data Center instances.

The interesting bit

The README ships documentation in llms.txt format—literally feeding its own docs back to LLMs—while supporting OAuth 2.0, personal access tokens, and multi-user HTTP transports. It is a thorough protocol translation layer, not a toy integration.

Key highlights

• 72 tools covering search, CRUD, and workflow transitions across Jira and Confluence
• Supports both Cloud and legacy Server/Data Center deployments (Confluence 6.0+, Jira 8.14+)
• Authentication flexibility: API tokens, PATs, or OAuth 2.0
• SSE and streamable HTTP transports for multi-user setups
• Documentation offered in machine-friendly llms.txt and llms-full.txt flavors

Caveats

• Not an official Atlassian product; the README makes this explicit
• Security section warns against sharing API tokens and exposing .env files

Verdict

Worth a look if your team lives in Jira/Confluence and you want an AI assistant to actually act on tickets rather than just summarize them. Skip if you aren’t already running an MCP client like Claude Desktop or Cursor.