# Cua - The Computer Use Agent Platform

> Open source infrastructure for computer-use agents

Cua - The Computer Use Agent Platform provides secure, isolated cloud containers specifically designed for computer-use AI agents. We make it easy to deploy and manage GUI environments for your AI agents without the infrastructure complexity. Skip the infrastructure headaches and run secure, isolated sandbox environments for your agents instantly.

**Backed by Y Combinator** - Supporting our mission to make AI agent infrastructure accessible to everyone.

## Key Links

- Website: https://cua.ai
- GitHub: https://github.com/trycua/cua
- Documentation: https://cua.ai/docs
- Blog: https://cua.ai/blog
- Pricing: https://cua.ai/pricing
- Developers: https://cua.ai/developers
- About: https://cua.ai/about

## API Documentation

API Swagger Docs: Contact support for API access
Base API URL: https://api.cua.ai

## Main Features

### Secure Agent Environments
- Deploy computer-use agents in isolated, secure cloud containers
- Sandbox environments for safe agent testing and execution
- GUI environments optimized for agents that control computers
- Support for Linux, Windows, and macOS agent workloads

### Infrastructure for Deploying Agents
- Provision agent containers in seconds without infrastructure complexity
- Multiple container sizes (small, medium, large, xlarge) for different agent workloads
- Real-time credit-based billing for cost-effective agent deployment
- Auto-scaling and resource management for production agent systems

### Developer Tools for Agent Development
- API keys for programmatic agent deployment and management
- Sandbox playground for testing computer-use agents
- Container support for Docker-based agent workflows
- Unified model provider access for vision-language model integration

### Security & Isolation
- Secure, isolated environments for each agent instance
- Enterprise SSO via Clerk for team-based agent management
- SSL certificate management for secure agent communication
- Protected sandbox execution prevents agent breakout

### Computer Use Stack

One unified interface for computer-use models. Execute powerful workflows with structured outputs, multi-turn conversations, and custom tools.

### Integration Options

**Agent SDK (Python)**
Build AI agents that automate workflows on Cua Computers.

```bash
pip install cua-agent
```

**MCP Server (Model Context Protocol)**
Run CUA through Claude Desktop or other MCP clients.

```bash
pip install cua-mcp-server
```

### Framework Integrations
- Anthropic Claude integration
- Model Context Protocol (MCP) support
- Compatible with major AI frameworks and LLMs
- REST API for custom integrations

## Sandbox Types

Available operating systems for computer-use agents:
- Linux (Ubuntu) - Full GUI support for Linux-based agents
- Windows - Native Windows environment for Windows automation agents
- macOS - macOS GUI for Apple ecosystem agents

Sandbox sizes optimized for different agent workloads:
- Small: Development and testing agents
- Medium: Production agents with moderate compute needs
- Large: Intensive computer-use tasks

## Pricing

### Open Source - Free
For developers and open-source projects who prefer self-hosting.

**Includes:**
- MIT License
- Windows, Linux, macOS local environments
- 100+ VLMs in Cua Agent Framework
- Community Discord support

Get started: https://github.com/trycua/cua

### Pro - $10+ (Most Popular)
Flexible credits for cloud compute and Computer-Use VLMs, with managed cloud environments and inference.

**Includes:**
- Cloud Linux and Windows Environments
- CUA LLM Inference Provider
- Cuazar Playground UI
- Priority support via Slack
- Custom credit amounts (minimum $10, up to $5000)

Purchase credits: https://cua.ai/pricing

### Enterprise - Custom Pricing
Scalable cloud sandboxes tailored for large teams and organizations.

**Includes:**
- Everything in Pro
- 24/7 support
- HIPAA compliance
- SOC Type 1/2 Reports
- Custom infrastructure solutions

Book a demo: https://cal.com/cua/cua-demo

## Getting Started with Agent Deployment

1. Sign up at https://cua.ai/signup
2. Purchase credits at https://cua.ai/pricing
3. Deploy your first secure agent from the dashboard
4. Connect your computer-use agent to the isolated environment
5. Run your agent safely in the sandbox without infrastructure setup

## Resources

- Blog posts with tutorials and updates: https://cua.ai/blog
- GitHub repository with examples: https://github.com/trycua/cua
- Hackathon resources: https://cua.ai/hackathon

## Use Cases

- **Computer-Use Agent Development**: Build and test agents that interact with GUIs
- **Autonomous Agent Deployment**: Deploy production computer-use agents at scale
- **Secure Agent Sandboxing**: Test agents in isolated environments before production
- **Multi-Agent Systems**: Run multiple computer-use agents in separate secure sandboxes
- **Agent Research**: Experiment with computer-use capabilities in controlled environments
- **Enterprise Agent Automation**: Deploy secure, managed agent infrastructure for teams

## Target Audience

- AI/ML researchers building computer-use agents
- Developers creating autonomous agents that control computers
- Companies deploying production computer-use agent systems
- Startups building agent-powered automation products
- Security teams testing agents in isolated sandbox environments
- Students and educators learning about computer-use AI

## Technology Stack

Open source infrastructure components:
- Frontend: React, TypeScript, React Router v7, Tailwind CSS v4
- Backend: Go (API service), Python (Resource Manager)
- Database: PostgreSQL (via Supabase)
- Infrastructure: Multi-cloud support (GCP, Azure, AWS)
- Auth: Clerk for enterprise authentication
- sandboxes: Docker-based secure agent isolation
- Open Source SDKs: Available on GitHub

## Frequently Asked Questions

### What is a Computer Use Agent and how can it help me?

A Computer Use Agent is an AI agent that interacts with your computer just like a human would. It can see your screen, move the cursor, click buttons, type text, and navigate between applications—all through visual understanding of the interface. Unlike traditional automation scripts that rely on rigid code paths, computer use agents use AI vision to understand what's on screen and take appropriate actions.

Cua.ai can automate complex workflows across any application: filling forms, managing files, conducting web research, interacting with desktop software, and executing multi-step tasks. It adapts to different interfaces dynamically, making it ideal for streamlining repetitive work and boosting productivity.

Get started: https://cua.ai/docs/cua/guide/get-started/what-is-cua

### What models can I use in my computer use agent?

Cua.ai supports almost all AI models and providers including Anthropic Claude, OpenAI, local models (Ollama, MLX, Hugging Face), and more.

Model providers documentation: https://cua.ai/docs/cua/guide/fundamentals/vlms

### Can Cua.ai automate tasks in any desktop environment?

Yes, Cua.ai can automate tasks in Linux, MacOS and Windows desktop environments through various options like Lume, Docker, cloud offering and more. Its flexibility allows it to interact with various applications, making it suitable for diverse automation needs.

Setup guide: https://cua.ai/docs/cua/guide/get-started/set-up-sandbox

### Is Cua.ai free?

Yes! Cua.ai is open source and free to use. You can self-host and run Cua.ai with your own infrastructure and AI models at no cost. We also offer cloud environments and enterprise tiers with additional features, support, and managed infrastructure.

Pricing details: https://cua.ai/pricing

## Policies

- Privacy Policy: https://cua.ai/privacy-policy
- Cookie Policy: https://cua.ai/cookie-policy