Docker Reader MCP Server

A fully self-sustaining Model Context Protocol (MCP) server that provides AI agents with seamless access to Docker container logs and information for debugging and monitoring purposes.

Features

🐳 Container Management

• List all containers (running and stopped)
• Inspect container details and configuration
• Get real-time resource usage statistics

📋 Log Access

• Read container logs with configurable options
• Support for timestamps, line limits, and time ranges
• Efficient log retrieval for debugging

🔧 AI-Friendly

• Simple integration with AI agents
• Structured data output
• Comprehensive error handling
• No user interaction required

Installation

Prerequisites

• Node.js 18+
• Docker installed and running
• Appropriate Docker permissions

Quick Setup

# Clone and install
git clone <repository>
cd docker-reader-mcp
npm install

# Build the server
npm run build

# Test the server
npm run dev

Usage with AI Agents

Configuration for Claude Code

Add to your MCP configuration file:

{
  "mcpServers": {
    "docker-reader": {
      "command": "node",
      "args": ["/path/to/docker-reader-mcp/dist/index.js"],
      "description": "Docker container log reader and inspector"
    }
  }
}

Django Development Example

For a Django project with multiple containers (django, celery, celery-beat, postgres, redis), the AI agent can now:

Agent: "Show me all running containers"
# Uses: docker_list_containers

Agent: "Get the last 50 lines of django container logs"
# Uses: docker_read_logs with container="django" and lines=50

Agent: "Check if there are any errors in celery logs from the last hour"
# Uses: docker_read_logs with container="celery" and since="1h"

Available Tools

docker_list_containers

Lists Docker containers with their status and details.

Parameters:

• all (boolean, optional): Include stopped containers (default: false)

Example:

{
  "name": "docker_list_containers",
  "arguments": {
    "all": true
  }
}

docker_read_logs

Retrieves logs from a specific container.

Parameters:

• container (string, required): Container ID or name
• lines (number, optional): Number of lines to retrieve (default: 100, max: 10000)
• since (string, optional): Show logs since timestamp (e.g., "1h", "30m", "2024-01-01T10:00:00Z")
• until (string, optional): Show logs before timestamp
• timestamps (boolean, optional): Include timestamps (default: false)

Example:

{
  "name": "docker_read_logs",
  "arguments": {
    "container": "my-django-app",
    "lines": 200,
    "since": "1h",
    "timestamps": true
  }
}

docker_inspect_container

Gets detailed information about a container.

Parameters:

• container (string, required): Container ID or name

Example:

{
  "name": "docker_inspect_container",
  "arguments": {
    "container": "postgres-db"
  }
}

docker_container_stats

Retrieves resource usage statistics for a running container.

Parameters:

• container (string, required): Container ID or name

Example:

{
  "name": "docker_container_stats",
  "arguments": {
    "container": "redis-cache"
  }
}

Resources

The server also exposes container information as MCP resources:

• docker://container/{container-id}: Detailed container inspection data

Error Handling

The server provides comprehensive error handling for:

• Docker not available: Clear message when Docker isn't installed/running
• Permission denied: Guidance on Docker permissions
• Container not found: Specific error when containers don't exist
• Invalid arguments: Detailed validation error messages
• Container not running: Helpful hints for accessing stopped containers

Architecture

src/
├── index.ts           # Main MCP server implementation
├── docker-client.ts   # Docker command wrapper and utilities
├── validation.ts      # Input validation schemas
└── errors.ts          # Custom error types and formatting

Key Design Principles

1. Self-Sustaining: No user interaction required during operation
2. Robust Error Handling: Comprehensive validation and error reporting
3. Efficient: Minimal overhead for log access and container queries
4. Secure: Input validation and safe Docker command execution
5. Modular: Clean separation of concerns for maintainability

Development

Scripts

npm run build      # Build TypeScript to JavaScript
npm run dev        # Run in development mode with tsx
npm run start      # Run the built server
npm run typecheck  # Type checking only
npm run lint       # Code linting

Testing

# Test with a running container
docker run -d --name test-container nginx
node dist/index.js

# In another terminal, test MCP calls
# (Use MCP client or testing tool)

Security Considerations

• The server only reads Docker information and logs
• No container manipulation capabilities
• Input validation prevents command injection
• Requires appropriate Docker permissions

Troubleshooting

Common Issues

"Docker is not available"

• Ensure Docker is installed and running
• Check if docker --version works in your terminal

"Permission denied"

• Add your user to the docker group: sudo usermod -aG docker $USER
• Or run with appropriate privileges

"Container not found"

• Use docker ps -a to list all containers
• Check container names and IDs
• Ensure you're using the correct container identifier

Contributing

1. Fork the repository
2. Create your feature branch
3. Add tests for new functionality
4. Ensure all tests pass
5. Submit a pull request

Contributors & Sponsors

This project is contributed to and sponsored by Debugg.ai - Advanced AI browser automation & testing tools

License

MIT License - see LICENSE file for details.

Made with 🩸, 💦, and 😭 in 🌉 San Francisco