This project implements a Model Context Protocol (MCP) server for interacting with Atlassian Confluence and Jira via LLMs.
- Get, update, and explore Confluence pages, folders, children, and spaces
- Perform advanced searches using CQL (Confluence Query Language)
- Interact with Jira issues (
create, update,delete, search) - Designed for LLM integration (e.g., Claude for Desktop)
Follow these steps to set up the MCP server locally:
- Node.js (v18 or later recommended)
- npm (comes with Node.js)
- API token for Atlassian. You can create on from https://id.atlassian.com/manage-profile/security/api-tokens
-
Clone the repository:
git clone https://github.com/Parthav46/atlassian-mcp.git cd atlassian-mcp
-
Install dependencies:
npm install
-
Configure environment:
- You can set configuration options using command-line arguments or environment variables:
--base-url <Confluence or Jira base URL>
--token <User token>
--username <Username>
- These arguments override environment variables if both are set.
- You can set configuration options using command-line arguments or environment variables:
-
Build the project:
npm run build
-
Run the server using npx:
npx atlassian-mcp --base-url <Confluence or Jira base URL> --token <User token> --username <Username>
- You can also use environment variables instead of command-line arguments.
Create MCP JSON config for Copilot or Claude for Desktop:
- Add the following content to configure your MCP server:
{ "command": "npx", "type": "stdio", "args": [ "atlassian-mcp@latest" ], "env": { "ATLASSIAN_BASE_URL": "<Your Jira Base URL>", "ATLASSIAN_API_TOKEN": "<Your Atlassian API Token>", "ATLASSIAN_USERNAME": "<Your Atlassian Username>" } }
- Replace the placeholders with your actual configuration values.
- This configuration works for both GitHub Copilot and Claude for Desktop.
-
npm run build
— Compile TypeScript -
npm start
— Run the server using compiled JS -
npm run dev
— Run the server with ts-node (for development)
This project is licensed under the MIT License. This project was created entirely using GitHub Copilot with GPT-4.1.