MCP 服务器

atlassian-mcp-server

An open-source Model Context Protocol (MCP) server for Atlassian Jira and Confluence Cloud, enabling LLMs to search, read, write, and manage issues and pages.

README

Atlassian MCP Server

A Model Context Protocol (MCP) server that provides powerful tools for interacting with Atlassian Jira and Confluence Cloud. Built with Python and the mcp SDK.

This server enables LLM agents (like Claude Desktop) to:

Jira: Search for issues using JQL, view issue details, add comments, transition issues, and create/update issues (with rich text support).
Confluence: Search space content, view pages, create new pages (including nested pages), and edit pages (with auto-versioning).

Features

Jira Tools

list_jira_issues: Search and list issues using JQL (Jira Query Language).
read_jira_issue: Retrieve full details of a specific issue.
jira_create_issue: Create new issues (Support for Projects, Issue Types, and ADF Descriptions).
jira_update_issue: Update issue summary and description.
jira_add_comment: Add comments to issues.
jira_get_comments: Retrieve all comments on an issue.
jira_download_attachment: Download an attachment (image/document) by its ID.
jira_transition_issue: Move issues through their workflow (e.g., To Do -> Done).

Confluence Tools

list_confluence_pages: List pages within a specific space.
view_confluence_page: Retrieve page content and metadata.
confluence_create_page: Create new pages, optionally nested under a parent page.
edit_confluence_page: Update page content (Automatically handles version increments).
- Note: Includes guidance for handling Mermaid diagrams via the Mermaid Diagrams plugin.
confluence_delete_page: Delete a Confluence page.
confluence_search: Perform advanced searches using CQL (Confluence Query Language).
confluence_get_comments: Retrieve all comments on a page.

Prerequisites

Python: Version 3.10+ (Tested with 3.14.2)
Atlassian Account: An Atlassian account with an API Token.

Setup

Clone and Install:

git clone https://github.com/tingyiy/atlassian-mcp-server.git
cd atlassian-mcp-server
pip install -r requirements.txt

Configuration: Create a .env file in the project root with your credentials:

ATLASSIAN_USERNAME=your_email@example.com
ATLASSIAN_API_KEY=your_api_token
JIRA_URL=https://your-domain.atlassian.net/rest/api/3
CONFLUENCE_URL=https://your-domain.atlassian.net/wiki
CONFLUENCE_SPACE_KEY=your_default_space_key

Note: JIRA_URL must point to the /rest/api/3 endpoint.

Run the Server:
```
python server.py
```

Usage

This MCP server can be used with any MCP-compliant client, including IDEs (like Cursor, VS Code) and desktop agents (like Claude Desktop).

Generic Configuration

Most MCP clients require the command to run the server and the environment variables.

Command: python /abs/path/to/atlassian-mcp-server/server.py
Environment: All variables from .env must be passed to the process.

Claude Desktop

Add the following to your claude_desktop_config.json:

{
  "mcpServers": {
    "atlassian": {
      "command": "/path/to/python",
      "args": [
        "/path/to/atlassian-mcp-server/server.py"
      ],
      "env": {
        "ATLASSIAN_USERNAME": "your_email@example.com",
        "ATLASSIAN_API_KEY": "your_api_token",
        "JIRA_URL": "https://your-domain.atlassian.net/rest/api/3",
        "CONFLUENCE_URL": "https://your-domain.atlassian.net/wiki",
        "CONFLUENCE_SPACE_KEY": "DS"
      }
    }
  }
}

Cursor

To use with Cursor (via the .cursor/mcp.json or settings):

{
  "mcpServers": {
    "namd": "atlassian",
    "command": "python3",
    "args": ["/path/to/atlassian-mcp-server/server.py"],
    "env": {
      "ATLASSIAN_USERNAME": "your_email@example.com",
      "ATLASSIAN_API_KEY": "your_api_token",
      "JIRA_URL": "https://your-domain.atlassian.net/rest/api/3",
      "CONFLUENCE_URL": "https://your-domain.atlassian.net/wiki"
    }
  }
}

Note: Ensure the python environment has the required dependencies installed.

Confluence Tips: Mermaid Diagrams

The MCP tools cannot render Mermaid diagrams programmatically because the standard plugin stores source code internally. To reference a diagram, provide the Mermaid source in a code block and ask the user to manually render it in the Confluence UI.

Example Agent Output for a Diagram: "I've added the following code block to the page. Please select it in the editor and convert it to a Mermaid diagram:"

<ac:structured-macro ac:name="code" ac:schema-version="1">
    <ac:parameter ac:name="language">text</ac:parameter>
    <ac:plain-text-body><![CDATA[graph TD; A-->B;]]></ac:plain-text-body>
</ac:structured-macro>