cua
Open-source infrastructure for training and evaluating AI agents capable of controlling full desktop environments.
- Region
- Overseas
- Pricing
- Free
- Open source
- Yes
- GitHub Stars
- ★ 17.6k
- Source
- GitHub
- Added
- 2026-06-04
- Last verified
- 2026-06-04
- GitHub
- github.com/trycua/cua
Overview
cua is an open-source infrastructure for training and evaluating AI agents that can control full desktop environments. It provides sandboxed environments, SDKs, and benchmarking tools to help developers build AI agents capable of autonomously completing tasks across operating systems such as macOS, Linux, and Windows. Through a unified API interface, these agents can be easily created and deployed, whether in cloud or local environments.
Key features
- ▪Multi-platform support: macOS, Linux, Windows
- ▪Sandboxed environment for development
- ▪Unified API simplifies cross-platform development
- ▪Includes benchmarking and reinforcement learning environments
Use cases
Pros
- ＋Excellent cross-platform compatibility
- ＋Easy integration with existing systems
- ＋Strong community support
Limitations / notes
- －Linux version is currently in pre-release stage
- －Requires some programming knowledge
Who it's for
This overview was compiled by AI from public sources and may contain inaccuracies — please refer to the official site.
FAQ
Which operating systems does cua support?
Currently supports macOS, Windows, and Linux (pre-release version).
How do I get started with cua?
Install the cua library via pip and follow the official documentation to build your first agent.
Similar agents
Cursor
FeaturedAI-powered code editor with Agent mode
Dify
FeaturedOpen-source LLMOps + Agent orchestration platform
扣子 Coze
FeaturedByteDance's no-code Agent/Bot building platform
Trae
FeaturedAI IDE (with Agent) from ByteDance
LangChain
FeaturedThe most mainstream LLM/Agent orchestration framework
Devin
FeaturedAutonomous software engineer Agent by Cognition
Something wrong? Let us know on the About page and we'll fix it.