mcp-atlassian

This MCP server integrates with Atlassian Cloud and Server/Data Center products to provide natural language tooling for Jira and Confluence. It exposes a set of Jira and Confluence tools (e.g., jira_search, jira_get_issue, confluence_search, confluence_get_page, and others) that let your AI assistant perform tasks like searching issues, creating or updating issues, and reading or creating Confluence pages. To use it, configure the MCP in your IDE or agent to point at the mcp-atlassian server and supply the required Atlassian tokens and URLs via environment variables. Once running, you can ask your assistant to perform actions such as finding issues, updating ticket statuses, or retrieving page content. The tools are organized by Jira and Confluence with clear commands, enabling multi-step workflows through natural language prompts.

Prerequisites:

• Python 3.8+ installed on your system
• Access to the internet to fetch dependencies
• Atlassian API tokens for Jira and Confluence (or PATs for Server/DC deployments)

Install and run:

1. Install the uvx launcher (the MCP runtime) so you can execute uvx commands:

   • pipx install uvx (or follow your environment’s preferred method to install uvx)

2. Install the MCP Atlassian package (this is invoked by the uvx launcher as shown in the configuration):

   • This is typically done by ensuring the package is available in your Python environment as mcp-atlassian on PyPI, or by using the uvx runner with the package name as shown in the mcp_config.

3. Verify installation:

   • uvx --version (or run the configured command to ensure the MCP server starts)

4. Prepare environment variables and start:

   • Use the provided mcp_config snippet to inject required environment variables (JIRA_URL, JIRA_USERNAME, JIRA_API_TOKEN, CONFLUENCE_URL, CONFLUENCE_USERNAME, CONFLUENCE_API_TOKEN) or JIRA_PERSONAL_TOKEN for Server/DC setups.
   • Start the MCP server via your IDE integration or by running the uvx command: uvx mcp-atlassian

5. Optional: configure your IDE or Claude Desktop to use the mcp-atlassian server by adding the snippet from the mcp_config example under mcpServers. Then you can interact with the tools through natural language prompts.

Environment variables map to Atlassian credentials required by the tools. For Server/Data Center deployments, use JIRA_PERSONAL_TOKEN in place of JIRA_USERNAME + JIRA_API_TOKEN. Keep tokens secure and do not commit them to source control. If you encounter authentication errors, re-issue tokens via Atlassian and update the environment variables accordingly. The Tools Reference contains the full list of Jira and Confluence commands (e.g., jira_search, confluence_get_page, jira_create_issue, confluence_create_page, etc.). Ensure URLs (JIRA_URL and CONFLUENCE_URL) point to the correct cloud or server endpoints for your deployment. If you need to run the server locally behind a proxy, configure proxy settings in your environment or via uvx configuration. For troubleshooting, consult the documentation at the provided docs site and the llms.txt resources for CLI and prompt usage guidance.