Atlassian Data Center MCP (Model Context Protocol)

Author: Stephen Chen

This project provides a Go-based Model Context Protocol (MCP) service for managing and interacting with Atlassian Data Center products including Jira, Confluence, and Bitbucket. It allows you to manage these products programmatically through a unified interface with configurable authentication and permissions.

Features

• Multi-Product Support: Unified interface for Jira, Confluence, and Bitbucket
• Model Context Protocol: Exposes tools through the Model Context Protocol for all operations
• Configuration Management: Flexible configuration through files, environment variables, and hot reloading
• Authentication: Direct API token authentication for each service
• Permissions: Fine-grained read/write permissions for each service
• Health Monitoring: Built-in health checks for all services
• Docker Support: Ready for containerized deployment
• Reverse Proxy Support: Configurations for Nginx and Traefik with HTTPS support

Running the Application

You can run the application in multiple ways: directly using Go, by building binaries first, or using Docker.

For detailed instructions on running the application:

• Running with Docker - Instructions for running the service using Docker and Docker Compose
• Running locally - Instructions for running the service directly on your system

Reverse Proxy Configuration

For reverse proxy configurations, please refer to the separate documentation:

• Nginx Configuration with mkcert - Instructions for setting up HTTPS access using Nginx
• Traefik Configuration with mkcert - Instructions for setting up HTTPS access using Traefik
• Production Deployment with Let's Encrypt - Instructions for production deployment with automatic SSL certificates

Below are basic commands for development purposes:

Direct Execution

# Run the server
go run cmd/server/main.go

# Run with custom config file path
go run cmd/server/main.go -c /path/to/your/config.yaml

Building and Running Binaries

The project uses a Makefile to simplify the build process. All binaries are placed in the dist directory.

# Show all available commands
make help

# Build the server binary for your current OS
make build

# Build a statically linked release binary
make release

After building the project, you can run the binary directly from the dist directory:

# Run the server
./dist/atlassian-dc-mcp-server

# Run with custom config file path
./dist/atlassian-dc-mcp-server -c /path/to/your/config.yaml

Configuration

The application requires a configuration file to operate. By default, it looks for config.yaml in the current directory, but you can specify a different path using the -c or --config flag.

A sample configuration file is provided as config.yaml.example. Copy this file to config.yaml and edit it with your settings:

cp config.yaml.example config.yaml

The configuration file is self-documented with examples for all available settings. Please refer to the config.yaml.example file for detailed configuration options.

Prune Configuration

The application includes a feature to remove sensitive or unnecessary fields from API responses before returning them to the client. This is configured through the prune section in the configuration file:

• fuzzy_keys: List of key prefixes that should be removed from responses
• remove_paths: List of exact paths or path suffixes that should be removed

Example:

prune:
  fuzzy_keys:
    - "customfield"
  remove_paths:
    - "emailAddress"
    - "avatarUrls"
    # ... other fields to remove

Authentication Modes

The service supports two authentication modes:

1. Config Mode (default): API tokens are read from the configuration file
2. Header Mode: API tokens are passed via HTTP headers

To enable header mode, start the server with the -auth-mode=header flag:

./dist/atlassian-dc-mcp-server -auth-mode=header

In header mode, the service expects the following HTTP headers:

• Jira-Token: API token for Jira
• Confluence-Token: API token for Confluence
• Bitbucket-Token: API token for Bitbucket

This mode is particularly useful when deploying the service in environments where you want to avoid storing sensitive tokens in configuration files, such as when using the service behind a reverse proxy that handles authentication.

Tools Documentation

Jira Tools

Tools for interacting with Jira:

• Get current user information
• Get issues
• Create issues
• And many more

Confluence Tools

Tools for interacting with Confluence:

• Get current user information
• Get content
• Search content
• And more

Bitbucket Tools

Tools for interacting with Bitbucket:

• Get current user information
• Get repositories
• Get commits
• And more

Lingma Rules

This project includes predefined Lingma rules that demonstrate how to use the Atlassian Data Center MCP service for automated code review tasks. For detailed information on how to use these rules, please refer to the Lingma Rules documentation.

The code_review.md rule file provides an example configuration showing how to leverage the Atlassian Data Center MCP service for automated Bitbucket PR code reviews. These rules help you:

• Automate code review processes using the MCP service
• Standardize code quality and security checks
• Improve interaction efficiency with AI assistants during code reviews

Development

Prerequisites

• Go 1.20 or higher
• Docker (for containerization)

Building

It is recommended to use the Makefile to build the project, which ensures all build artifacts are placed in the unified dist directory:

# Build server binary to dist directory
make build

To build binaries for specific operating systems:

# Build for Linux
make build-linux

# Build for Windows
make build-windows

# Build for macOS
make build-macos

Benefits of using make commands:

• All build artifacts are placed in the dist directory
• Automatically handles cross-platform builds
• Ensures consistent build parameters

Contributing

Contributions are welcome! Please feel free to submit a Pull Request.

License

This project is licensed under the MIT License - see the LICENSE file for details.

References

• Confluence REST API
• Jira REST API
• Bitbucket REST API
• Model Context Protocol Go SDK