Atlassian MCP Server

Atlassian MCP Server

The Atlassian MCP Server is an open-source, standardized integration server that connects AI assistants and IDEs to Atlassian products like Jira and Confluence, enabling seamless, secure, and flexible access to Atlassian data and actions through MCP-compatible applications.

Author: sooperset


View Protocol

What is Atlassian MCP Server?

Atlassian MCP Server is a Model Context Protocol (MCP) server designed to integrate Atlassian products (Confluence and Jira, both Cloud and Server/Data Center) with AI-powered assistants and developer tools. By exposing Jira and Confluence capabilities as executable functions ("tools") over the MCP interface, it allows LLMs and agents to automate, search, and interact with Atlassian content programmatically. It acts as a secure bridge between your Atlassian instance and MCP host applications.

How to Configure Atlassian MCP Server

  1. Authentication Setup: Choose your preferred authentication method, based on deployment type:
    • For Jira/Confluence Cloud: Use API Token (recommended) or OAuth 2.0
    • For Server/Data Center: Use Personal Access Token (PAT)
  2. Environment Variables: Prepare environment variables for your Atlassian URLs, tokens, and other options (such as project/space filters, read-only mode, logging).
  3. Installation: Pull the Docker image (ghcr.io/sooperset/mcp-atlassian:latest) to your local environment or server.
  4. Configuration in IDE or Assistant:
    • Directly pass environment variables via your IDE’s MCP server settings (e.g., in claude_desktop_config.json or Cursor’s MCP settings).
    • Alternatively, create an environment file and use Docker's --env-file flag.
  5. Transport Setup: The server can run as a process using stdio, or as a persistent HTTP service (using SSE or streamable HTTP transports) for multi-user or networked deployments.
  6. Proxy Support: Configure HTTP(S)/SOCKS proxies as needed using the corresponding environment variables.
  7. Tool Control: Use ENABLED_TOOLS to restrict or allow access to specific operations, and READ_ONLY_MODE for further security.

How to Use Atlassian MCP Server

  1. Start the Server: Launch the server with your configuration via Docker, optionally specifying enabled tools, logging, or transport type.
  2. Connect via IDE or AI Assistant: Add the running server as an MCP server in your IDE/assistant settings by providing the appropriate command or HTTP endpoint.
  3. Invoke Tools: Use natural language or custom flows to ask your assistant for help with Jira and Confluence actions (e.g., search, create, update, summarize).
  4. Control Access: Filter available tools, enable/disable write operations, and choose per-user or server-wide authentication as needed.
  5. Monitor and Debug: View logs, use debugging tools, and monitor interactions to ensure smooth operation.

Key Features

  • Supports Jira & Confluence (Cloud and Server/DC) with a single integration
  • Executable Tools: Exposes Atlassian APIs as invokable tools via MCP for use by LLMs and agents
  • Seamless IDE/Assistant Integration: Configurable in Claude Desktop, Cursor, and similar platforms
  • Multi-Auth Support: Supports OAuth 2.0, API tokens, and PATs. Enables both single and multi-user authentication
  • Secure In-Infra Deployment: Keeps credentials and data inside your environment
  • Tool Filtering & Read-Only Modes: Fine-grained control over available actions and permissions
  • Proxy and Logging Support: Enterprise-ready with custom proxy, logging, and verbose debug options
  • Scalable HTTP Transport: Can run as a persistent HTTP service for shared or automated use cases

Use Cases

  • AI-Powered Search: Quickly find and summarize project documentation and issues using natural language queries
  • Automated Jira Updates: Create, update, transition, and comment on Jira issues based on meeting notes or chat commands
  • Document Generation: Create and update Confluence pages or Jira tickets directly from your AI assistant or IDE
  • Reporting and Filtering: Automate fetching and summarization of urgent bugs, status updates, and project metrics
  • Multi-User Workflows: Enable organization-wide access to Atlassian actions, respecting each user's own permissions using per-user tokens

FAQ

Q: What authentication methods are supported?
A: The server supports Atlassian API Tokens (Cloud), Personal Access Tokens (Server/DC), and OAuth 2.0 (Cloud, for advanced scenarios and multi-user support).

Q: Can I restrict which actions the MCP server provides to my assistant?
A: Yes, you can specify enabled tools via the ENABLED_TOOLS environment variable or command-line flag, and set READ_ONLY_MODE to disable all write operations.

Q: How do I handle SSL certificate issues on my Server/Data Center deployment?
A: Set CONFLUENCE_SSL_VERIFY=false or JIRA_SSL_VERIFY=false in your configuration if your Atlassian instance uses self-signed certificates.

Q: Is it possible to run the server for multiple users with their own credentials?
A: Yes, use the HTTP transport (SSE or streamable HTTP) and have each client provide user-specific tokens (OAuth for Cloud or PAT for Server/DC).

Q: What if my assistant can’t connect to Jira or Confluence?
A: Check authentication credentials, network/proxy settings, URL accuracy, and user permissions in your Atlassian instance.