Atlassian

The MCP Atlassian server is a Model Context Protocol server that connects your AI assistant directly to your Atlassian products. It supports both Confluence and Jira, and it works with Cloud, Server, and Data Center versions.

This MCP server lets you ask your AI to perform tasks like updating Jira tickets from meeting notes, searching for technical documents in Confluence, or creating new issues without ever leaving your chat interface. It turns your AI into a junior team member who can handle all your Atlassian busywork.

Features

• 🔐 Multiple authentication methods (API tokens, PAT, OAuth 2.0)
• ☁️ Full support for both Cloud and Server/Data Center deployments
• 🔧 Configurable tool filtering and read-only mode
• 🌐 HTTP transport support for multi-user scenarios
• 📝 Custom headers and proxy configuration for enterprise environments
• ⚡ Docker-based deployment for easy setup
• 🔍 Comprehensive search and content management tools

Use Cases

• Streamline Meeting Follow-ups: After a sprint planning meeting, you can paste the notes into your AI and say, “Create Jira tickets for these action items.” The server handles the creation of each issue, assigning them and filling in the details, which saves a significant amount of manual entry.
• Instant Access to Documentation: When you’re coding and need to reference an architectural decision, you can ask, “Find the tech design doc for the payment gateway in Confluence and summarize it.” You get the information you need without breaking your focus by switching contexts to a web browser.
• Efficient Bug Reporting and Triage: You spot a bug and need to report it quickly. Just tell your AI, “Create an urgent bug in the ‘PROJ’ project titled ‘User login fails’ and add these replication steps.” You can also quickly check on progress by asking, “Show me all high-priority bugs in the PROJ project from the last week.”

How to Use It

You need Docker installed to run the MCP Atlassian server. The entire process involves setting up authentication, pulling the Docker image, and configuring your IDE.

1. You have three ways to authenticate with your Atlassian products.

• API Token (Cloud): This is the recommended method for Jira and Confluence Cloud. Go to your Atlassian profile’s security settings, create an API token, and copy it.
• Personal Access Token (Server/Data Center): For self-hosted instances, navigate to your profile and create a Personal Access Token (PAT).
• OAuth 2.0 (Cloud): This is a more complex but secure method for Cloud instances. It requires setting up an OAuth 2.0 app in the Atlassian Developer Console and running a setup wizard. It is best for advanced applications or multi-user scenarios.

2. Pull the pre-built Docker image from the GitHub container registry.

docker pull ghcr.io/sooperset/mcp-atlassian:latest

3. Tell your AI assistant how to run the server. This is done by adding a configuration object to your IDE’s settings file (e.g., for Claude Desktop or Cursor).

Below is a standard configuration for Atlassian Cloud using API tokens. You pass your credentials as environment variables directly in the configuration.

{
  "mcpServers": {
    "mcp-atlassian": {
      "command": "docker",
      "args": [
        "run",
        "-i",
        "--rm",
        "-e", "CONFLUENCE_URL",
        "-e", "CONFLUENCE_USERNAME",
        "-e", "CONFLUENCE_API_TOKEN",
        "-e", "JIRA_URL",
        "-e", "JIRA_USERNAME",
        "-e", "JIRA_API_TOKEN",
        "ghcr.io/sooperset/mcp-atlassian:latest"
      ],
      "env": {
        "CONFLUENCE_URL": "https://your-company.atlassian.net/wiki",
        "CONFLUENCE_USERNAME": "[email protected]",
        "CONFLUENCE_API_TOKEN": "your_confluence_api_token",
        "JIRA_URL": "https://your-company.atlassian.net",
        "JIRA_USERNAME": "[email protected]",
        "JIRA_API_TOKEN": "your_jira_api_token"
      }
    }
  }
}

4. For Server or Data Center instances, you would use Personal Access Tokens instead.

{
  "mcpServers": {
    "mcp-atlassian": {
      "command": "docker",
      "args": [
        "run",
        "--rm",
        "-i",
        "-e", "CONFLUENCE_URL",
        "-e", "CONFLUENCE_PERSONAL_TOKEN",
        "-e", "JIRA_URL",
        "-e", "JIRA_PERSONAL_TOKEN",
        "ghcr.io/sooperset/mcp-atlassian:latest"
      ],
      "env": {
        "CONFLUENCE_URL": "https://confluence.your-company.com",
        "CONFLUENCE_PERSONAL_TOKEN": "your_confluence_pat",
        "JIRA_URL": "https://jira.your-company.com",
        "JIRA_PERSONAL_TOKEN": "your_jira_pat"
      }
    }
  }
}

5. The MCP server also supports more advanced setups, including using an environment file (--env-file), proxy configurations, and running as a persistent HTTP service for multi-user applications. Check out the official documentation for more details.

Jira Tools

Read Operations

• jira_search: Find issues using a Jira Query Language (JQL) string.
• jira_get_issue: Retrieve the complete details for a single issue by its key.
• jira_get_all_projects: List all projects accessible to the user.
• jira_get_project_issues: Fetch all issues within a specific project.
• jira_get_worklog: Get all worklog entries for a particular issue.
• jira_get_transitions: List the possible workflow statuses an issue can be moved to.
• jira_search_fields: Find available fields that can be used in issues.
• jira_get_agile_boards: Get a list of all agile boards.
• jira_get_board_issues: Retrieve all issues belonging to a specific agile board.
• jira_get_sprints_from_board: List all sprints associated with a given board.
• jira_get_sprint_issues: Fetch all issues assigned to a specific sprint.
• jira_get_issue_link_types: Get all valid types of links that can exist between issues.
• jira_batch_get_changelogs: Retrieve the history of changes for multiple issues at once (Cloud only).
• jira_get_user_profile: Look up a user’s profile details.
• jira_download_attachments: Download files attached to an issue.
• jira_get_project_versions: List all the defined versions for a project.

Write Operations

• jira_create_issue: Create a new Jira issue.
• jira_update_issue: Modify fields on an existing issue.
• jira_delete_issue: Permanently remove an issue.
• jira_batch_create_issues: Create multiple new issues in a single operation.
• jira_add_comment: Post a new comment on an issue.
• jira_transition_issue: Change the status of an issue (e.g., from “To Do” to “In Progress”).
• jira_add_worklog: Log time spent working on an issue.
• jira_link_to_epic: Assign an issue to a specific epic.
• jira_create_sprint: Create a new sprint on a board.
• jira_update_sprint: Modify the details of an existing sprint.
• jira_create_issue_link: Create a link between two issues (e.g., “duplicates” or “is blocked by”).
• jira_remove_issue_link: Delete a link between two issues.
• jira_create_version: Create a new version for a project.
• jira_batch_create_versions: Create multiple new project versions at once.

Confluence Tools

Read Operations

• confluence_search: Find pages, blog posts, or comments using Confluence Query Language (CQL).
• confluence_get_page: Retrieve the full content and details of a specific page.
• confluence_get_page_children: List all the direct child pages under a parent page.
• confluence_get_comments: Fetch all comments from a given page.
• confluence_get_labels: Get the labels attached to a specific page.
• confluence_search_user: Find a Confluence user.

Write Operations

• confluence_create_page: Create a new Confluence page.
• confluence_update_page: Modify the content or title of an existing page.
• confluence_delete_page: Permanently remove a page.
• confluence_add_label: Add one or more labels to a page.
• confluence_add_comment: Post a new comment on a page.

FAQs

Q: My connection to our on-premise Jira Server is failing with an SSL error. How do I fix it?
A: This typically happens if your company uses self-signed SSL certificates. You can bypass verification by setting the JIRA_SSL_VERIFY environment variable to "false" in your IDE configuration. The same applies to Confluence with CONFLUENCE_SSL_VERIFY.

Q: Can I prevent the AI from making changes, like creating or deleting Jira tickets?
A: Yes. You can run the server in read-only mode by setting the READ_ONLY_MODE environment variable to "true". This will disable all tools that perform write operations.

Q: I only use Jira. Do I need to provide Confluence credentials?
A: No. The server is modular. You can provide credentials for only the service you intend to use. Just remove the Confluence-related environment variables from the args and env sections of your configuration.

Q: The server isn’t finding my project. What should I check first?
A: First, confirm that the account associated with your API token or PAT has the necessary permissions to view that project in Jira. Second, check if you have set the JIRA_PROJECTS_FILTER environment variable, which restricts the server to specific project keys. If that variable is set, make sure your project’s key is included in the comma-separated list.