MCP-DOCKER-Atlassian

Docker-based setup for the MCP Atlassian integration to use with VS Code's Copilot MCP extension.

This repository contains configuration and scripts to easily run the MCP Atlassian server in a Docker container, providing access to Confluence and Jira resources via the Model Context Protocol.

Note: Previous MyPy type checking issues inherited from the upstream project have been resolved. The type checking system is now fully functional and passing all checks.

Prerequisites

• Docker installed and running
• Visual Studio Code with the Copilot Chat extension
• Atlassian Cloud account with API token
  • Generate token at: https://id.atlassian.net/manage-profile/security/api-tokens
  • Save the token securely - it cannot be viewed again after creation

Quick Start

1. Clone the repository:

   git clone https://github.com/shane-reaume/MCP-DOCKER-Atlassian.git
   cd MCP-DOCKER-Atlassian
   

2. Set up your credentials: Copy the example environment file and edit it:

   cp .env.example .env
   # Edit .env with your Atlassian credentials
   

   Important: Do not use quotes around values in the .env file

3. Make scripts executable:

   chmod +x *.sh
   

4. Start the MCP server:

   ./start-mcp.sh
   

   The server will be available at http://localhost:9001/sse Stop the MCP server:

   ./stop-mcp.sh
   

5. Configure VS Code: Add this to your VS Code settings.json:

   "mcpManager.servers": [
     {
       "name": "Atlassian-MCP",
       "type": "sse",
       "url": "http://localhost:9001/sse",
       "enabled": true
     }
   ]
   

6. Stop the server when done:

   ./stop-mcp.sh
   

Available Scripts

• start-mcp.sh: Builds the Docker image and starts the container
• stop-mcp.sh: Stops the running container
• check-mcp.sh: Diagnostic tool to verify the server is running correctly
• cleanup.sh: Removes old containers and images
• update-remote.sh: Helper for managing git remotes
• mcp-manager.sh: Additional management commands

Available MCP Commands

Once connected, you can use commands like:

• @mcp confluence_search "search term"
• @mcp confluence_get_page "Page Title"
• @mcp jira_get_issue "PROJECT-123"
• @mcp jira_search "project = PROJECT AND status = 'In Progress'"

Troubleshooting

1. SSE Connection Issues

   • Ensure the Docker container is running: docker ps | grep mcp-atlassian
   • Check container logs: docker logs mcp-atlassian
   • Verify the port is accessible: curl http://localhost:9001/health

2. Authentication Issues

   • Ensure your API token is correct and not surrounded by quotes in .env
   • Check that your email matches the one used to generate the API token
   • Verify you have appropriate permissions in Confluence/Jira

3. Common Errors

   • "SSE error: TypeError: fetch failed": Restart VS Code and the MCP server
   • "Current user not permitted": Check API token and permissions
   • Empty results: Verify space/project access and try more general searches first

Environment Variables

Key environment variables in .env:

• MCP_TRANSPORT: Set to 'sse' for VS Code integration
• MCP_PORT: Default 9001, change if port is in use
• CONFLUENCE_URL: Your Atlassian instance URL
• CONFLUENCE_USERNAME: Your Atlassian email
• CONFLUENCE_API_TOKEN: Your API token
• CONFLUENCE_SPACES_FILTER: Optional space restrictions

See .env.example for all available options.

Based On

This setup is based on the MCP-Atlassian project, customized for Docker-based deployment.