AgentDock

AgentDock

【中文 | English】

Dynamic MCP Container Orchestration Platform

A Docker-based container orchestration system for deploying and managing
MCP (Model Context Protocol) services with dynamic lifecycle management.

---

✨ Features

• 🐳 Dynamic Container Management - Auto create, start, stop, and remove Docker containers
• 📊 Resource Monitoring - Real-time CPU, memory, and disk usage monitoring
• 🔄 Health Checks - Automatic container health detection with auto-restart
• 🌐 MCP Routing - Streamable-HTTP MCP protocol request routing
• 📈 Web Dashboard - Visual container management interface

🏗️ Architecture

AgentDock/
├── master/                     # Manager service
│   ├── main.py                 # FastAPI main application
│   ├── config.py               # Configuration management
│   ├── node.py                 # Node management routes
│   ├── config.yml              # Default configuration
│   ├── dockerfile              # Docker build file
│   └── templates/              # Web templates
├── node/                       # Base node image
├── agentdock-node-full/        # Full-feature MCP node
├── agentdock-node-explore/     # Explore MCP node (search & analysis)
├── docker-compose.yml          # Docker Compose configuration
└── .env.example                # Environment variables template

🚀 Quick Start

Option 1: Pull from Docker Hub (Recommended)

The fastest way to get started - no build required!

1. Configure Environment

# Clone the repository (or just download the compose file)
git clone https://github.com/OpenBMB/AgentCPM.git
cd AgentCPM/AgentCPM-Explore/AgentDock

# Create environment file
cp .env.example .env
# Edit .env file with your MongoDB credentials

2. Pull and Start Services

# Pull images from Docker Hub
docker compose -f docker-compose-hub.yml pull

# Start all services
docker compose -f docker-compose-hub.yml up -d

3. Access Dashboard

Open browser: http://localhost:8080

Docker Hub Images

Image	Description	Architectures
sailaoda/agentdock-manager	Main orchestration service	amd64, arm64
sailaoda/agentdock-node-full	Full-feature MCP server	amd64, arm64
sailaoda/agentdock-node-explore	Explore MCP server	amd64, arm64

# Pull individual images
docker pull sailaoda/agentdock-manager:latest
docker pull sailaoda/agentdock-node-full:latest
docker pull sailaoda/agentdock-node-explore:latest

---

Option 2: Build from Source

Build images locally from source code.

🌍 Global Build (Default)

For users outside China, use the default Dockerfiles with official mirrors:

# 1. Configure environment
cp .env.example .env
# Edit .env file with your MongoDB credentials

# 2. Build and start services
docker compose build
docker compose up -d

🇨🇳 China Build (Aliyun Mirrors)

For users in China, use the .cn Dockerfiles with Aliyun mirrors for faster builds:

# 1. Configure environment
cp .env.example .env
# Edit .env file with your MongoDB credentials

# 2. Build and start services (using China mirrors)
docker compose -f docker-compose.cn.yml build
docker compose -f docker-compose.cn.yml up -d

Files for China Build:

• master/dockerfile.cn
• agentdock-node-full/dockerfile.cn
• agentdock-node-explore/Dockerfile.cn
• docker-compose.cn.yml

3. Access Dashboard

Open browser: http://localhost:8080

📦 Services

Service	Description	Ports
agentdock-manager	Main orchestration dashboard	8080
agentdock-mongodb	Database for node persistence	27017
agentdock-node-full	Full-feature MCP server	8004, 8092
agentdock-node-explore	Explore MCP server (search & analysis)	8014, 8102

⚙️ Configuration

Environment Variables

Variable	Description	Required
MONGODB_USERNAME	MongoDB username	✅
MONGODB_PASSWORD	MongoDB password	✅
JINA_API_KEY	Jina Reader API key	❌
GOOGLE_SERP_API_KEY	Google SERP API key	❌

Resource Limits

Default resource allocation:

• agentdock-manager: 2 cores / 4GB
• agentdock-mongodb: 2 cores / 6GB
• agentdock-node-full: 8 cores / 32GB
• agentdock-node-search: 4 cores / 16GB

Adjust in docker-compose.yml as needed.

🔌 API Endpoints

Core Endpoints

Endpoint	Method	Description
/	GET	Web dashboard
/alive	GET	Health check
/node/create	POST	Create new node
/node/close	POST	Stop node
/node/release	POST	Release node
/node/details	GET	Get node details
/container/{id}/mcp	POST	MCP request routing

Resource Monitoring

Endpoint	Method	Description
/api/resources/system	GET	System resources
/api/resources/docker	GET	Docker container info
/api/resources/alerts	GET	Resource alerts

🛠️ MCP Operations

Create MCP Node

curl -X POST "http://localhost:8080/node/create?image_name=agentdock-node:explore"

Call MCP Tool

curl -X POST "http://localhost:8080/container/{container_id}/mcp" \
  -H "Content-Type: application/json" \
  -d '{
    "jsonrpc": "2.0",
    "method": "tools/call",
    "params": {
      "name": "tool_name",
      "arguments": {}
    },
    "id": 1
  }'

🔧 Development

Local Development

# Install dependencies
pip install -r master/requirements.txt

# Start development server
cd master
uvicorn main:app --reload --host 0.0.0.0 --port 8080

Build Images

# Build all images
docker compose build

# Rebuild without cache
docker compose build --no-cache

Common Commands

# Start services
docker compose up -d

# Stop services
docker compose down

# View logs
docker compose logs -f agentdock-manager

# Check status
docker compose ps

# Resource stats
docker stats

📄 License

The code in this repository is released under the Apache-2.0 license.