Content
# Atlassian MCP Server
A read-only Model Context Protocol (MCP) server for Atlassian products (Confluence and Jira).
## Installation
```bash
pip install -e .
```
## Configuration
Set your Atlassian credentials as environment variables:
```bash
export ATLASSIAN_DOMAIN=your-company.atlassian.net
export ATLASSIAN_EMAIL=your-email@company.com
# Separate tokens for each service (recommended)
export ATLASSIAN_CONFLUENCE_TOKEN=your-confluence-token
export ATLASSIAN_JIRA_TOKEN=your-jira-token
# OR for backward compatibility, single token (if it has access to both)
export ATLASSIAN_API_TOKEN=your-api-token
```
### Creating API Tokens
1. Go to: https://id.atlassian.com/manage-profile/security/api-tokens
2. Click "Create API token"
**For Confluence:**
- When creating a Confluence token, select these scopes:
- `read:content:confluence` - View pages, blog posts, comments, attachments
- `read:content-details:confluence` - View content metadata and properties
**For Jira:**
- When creating a Jira token, you'll see different scope options. Look for:
- Scopes that allow **reading issues** and **projects**
- **Search** capabilities for JQL queries
- Basic **browse** permissions
> **Note:** The exact scope names may vary in the Atlassian interface. Choose the minimal read-only scopes that allow viewing issues, projects, and searching. If you're unsure, you can start with broader read permissions and restrict them later.
## Usage
Start the MCP server:
```bash
atlassian-mcp
```
## Available Tools
### Confluence
- **confluence_get_page** - Get a page by ID
- **confluence_get_page_by_url** - Get a page by URL
- **confluence_search_pages** - Search for pages
- **confluence_list_spaces** - List all spaces
### Jira
- **jira_get_issue** - Get an issue by key (e.g., PROJ-123)
- **jira_get_issue_by_url** - Get an issue by URL
- **jira_search_issues** - Search issues using JQL
- **jira_list_projects** - List all projects
## Supported URLs
### Confluence Pages
- `https://domain.atlassian.net/wiki/spaces/SPACE/pages/123456/Page+Title`
- `https://domain.atlassian.net/wiki/display/SPACE/Page+Title?pageId=123456`
- `https://domain.atlassian.net/wiki/pages/viewpage.action?pageId=123456`
### Jira Issues
- `https://domain.atlassian.net/browse/PROJ-123`
- `https://domain.atlassian.net/jira/software/projects/PROJ/boards/1?selectedIssue=PROJ-123`
## MCP Client Configuration
```json
{
"mcpServers": {
"atlassian": {
"command": "python",
"args": ["-m", "atlassian_mcp.server"],
"cwd": "/path/to/your/atlassian_mcp",
"env": {
"ATLASSIAN_DOMAIN": "your-company.atlassian.net",
"ATLASSIAN_EMAIL": "your-email@company.com",
"ATLASSIAN_CONFLUENCE_TOKEN": "your-confluence-token",
"ATLASSIAN_JIRA_TOKEN": "your-jira-token"
}
}
}
}
```
### API Endpoints Used
This server uses the following Atlassian REST API endpoints:
**Confluence REST API:**
- `GET /wiki/rest/api/content/{id}` - Get page content
- `GET /wiki/rest/api/content/search` - Search pages using CQL
- `GET /wiki/rest/api/space` - List spaces
**Jira REST API v3:**
- `GET /rest/api/3/issue/{issueIdOrKey}` - Get issue details
- `GET /rest/api/3/search` - Search issues using JQL
- `GET /rest/api/3/project` - List projects
All endpoints are read-only and require basic authentication with your email and API token.
Connection Info
You Might Also Like
markitdown
MarkItDown-MCP is a lightweight server for converting URIs to Markdown.
markitdown
Python tool for converting files and office documents to Markdown.
Filesystem
Node.js MCP Server for filesystem operations with dynamic access control.
Sequential Thinking
A structured MCP server for dynamic problem-solving and reflective thinking.
Fetch
Retrieve and process content from web pages by converting HTML into markdown format.
TrendRadar
TrendRadar: Your hotspot assistant for real news in just 30 seconds.