🥩 Meatloaf — AI Agent Sandbox Runtime

Build. Run. Test. Ship. 🚀

📦 Install from npm: npm install -g meatloaf-cli

🔗 npm: https://www.npmjs.com/package/meatloaf-cli

🔗 GitHub: https://github.com/Atum246/meatloaf

Meatloaf is an open-source CLI tool that gives AI agents the power to not just write code, but to build, run, test, and ship applications autonomously. It bridges the gap between "generating code" and "delivering working software."

🤔 Why Meatloaf?

AI agents can write thousands of lines of code, but they can't:

• 🖥️ Spin up and interact with running services
• 👁️ See what the application actually looks like
• 🧪 Test endpoints and verify behavior
• 🐛 Debug issues in real-time
• 🔄 Iterate on build→test→fix loops

Meatloaf changes that. It gives AI agents a body to interact with the digital world.

⚡ Quick Start

# Install
npm install -g meatloaf-cli

# Create a sandbox
meatloaf create --image node:20-alpine

# Run commands
meatloaf exec sandbox-abc123 "npm init -y && npm install express"

# Test endpoints
meatloaf test --url http://localhost:3000

# Take screenshots
meatloaf screenshot http://localhost:3000

# Full lifecycle
meatloaf up ./my-project

🎯 Commands

Sandbox Management

Execution

HTTP Testing

Observation

Network Management

Volume Management

Image Management

Multi-Container Orchestration

Process Management

Filesystem Operations

Snapshots

Health & Monitoring

Port Management

Secrets Management

Environment Profiles

Git Operations

File Sync

Proxy & Tunneling

Scheduled Tasks

Diff & Compare

Watch Mode

Resource Quotas

Service Discovery

Audit Logging

System

🤖 AI Agent Integration

Meatloaf works seamlessly with:

OpenClaw

# Meatloaf works out of the box with OpenClaw agents
meatloaf create --image node:20-alpine
meatloaf exec sandbox-id "npm start"
meatloaf test --url http://localhost:3000

Claude Code / Cursor / VS Code

# Use meatloaf from any terminal-based AI agent
meatloaf up ./my-project --json  # JSON output for parsing

Any AI Agent

# Detect environment
meatloaf agent --detect

# JSON output mode
meatloaf create --image python:3.12-slim --json

📁 Templates

meatloaf templates           # List all
meatloaf init -t node-express
meatloaf init -t python-flask
meatloaf init -t static-site
meatloaf init -t go-gin
meatloaf init -t postgres
meatloaf init -t redis

🔥 Full Lifecycle Example

# One command to rule them all
meatloaf up ./my-app

# This will:
# 1. Create a sandbox
# 2. Upload your project files
# 3. Run setup commands (npm install, etc.)
# 4. Start the application
# 5. Wait for the endpoint to be ready
# 6. Report success

🧪 Batch Testing

Create a tests.json:

[
  { "name": "Home", "url": "http://localhost:3000", "method": "GET" },
  { "name": "Health", "url": "http://localhost:3000/health", "method": "GET" },
  { "name": "Create", "url": "http://localhost:3000/api/items", "method": "POST", "body": {"name": "test"} }
]

Run:

meatloaf batch tests.json

⚙️ Configuration

Meatloaf stores config in ~/.meatloaf/config.json:

{
  "defaultImage": "node:20-alpine",
  "sandboxTTL": 3600000,
  "maxSandboxes": 10,
  "screenshot": {
    "width": 1280,
    "height": 720
  }
}

🏗️ Architecture

meatloaf/
├── bin/meatloaf.js          # Entry point
├── src/
│   ├── index.js             # CLI definition
│   ├── commands/
│   │   ├── sandbox.js       # Sandbox lifecycle
│   │   ├── exec.js          # Command execution
│   │   ├── http.js          # HTTP testing
│   │   ├── screenshot.js    # Screenshot capture
│   │   ├── logs.js          # Log management
│   │   └── init.js          # Project templates
│   ├── lib/
│   │   ├── docker.js        # Docker engine
│   │   ├── http.js          # HTTP client
│   │   └── screenshot.js    # Screenshot engine
│   └── utils/
│       ├── logger.js        # CLI output
│       ├── config.js        # Configuration
│       └── helpers.js       # Utilities
└── tests/
    └── integration.test.js  # Test suite

📄 License

MIT — Use it, fork it, make it yours. 🥩

Built with ❤️ for AI agents everywhere 🤖