cua-agent
Cua (Computer Use) Agent for AI-driven computer interaction
What it is and what it does
Cua Agent is a framework for building AI-driven agents that can interact with computers—running commands, manipulating interfaces, and automating workflows across macOS, Windows, and Linux. It integrates with liteLLM to support multiple AI model providers, allowing developers to define agentic workflows that execute in sandboxed environments. The package handles the complexity of cross-platform computer interaction, providing a structured API for agents to perceive and act on desktop systems.
The framework is designed for scenarios where an AI model needs to autonomously perform tasks on a computer—from testing and quality assurance to data extraction and system administration. It abstracts away platform-specific details while maintaining security through sandboxing, and its integration with liteLLM means you can swap between different AI providers without rewriting agent logic.
Use it for:
- Automate repetitive desktop tasks like form filling, data entry, or file management across platforms
- Build AI-powered testing agents that interact with web applications and desktop software
- Create autonomous workflows for system administration, monitoring, and configuration tasks
- Develop agents that extract information from visual interfaces and unstructured desktop data
- Prototype and test agentic AI behaviors in controlled sandbox environments before deployment
Worth the install?
AI-flagged interpretation of the facts on this page — verify before relying
Cua Agent provides a framework for building agentic workflows that interact with computers through AI models, supporting macOS, Windows, and Linux sandboxes with liteLLM integration.
Yes, if you need to build AI agents that interact with computers across multiple platforms. The low install friction, active maintenance, and integration with liteLLM make it accessible. However, verify the license terms before use—the unclear license treatment is a blocker for some contexts. No known vulnerabilities and solid dependency choices support adoption.
Install
cua-agent on PyPI
pip
pip install cua-agentuv
uv add cua-agentpoetry
poetry add cua-agentInstalling cua-agent
Before you install
Low install friction with a pure-Python wheel distribution. Active maintenance status with a recent release cycle (51 days since last update). Twelve runtime dependencies including established libraries like httpx, aiohttp, pydantic, and litellm suggest a mature dependency ecosystem.
License in practice
License treatment is unclear—no SPDX identifier or raw license text is available in the package metadata. Verify the actual license terms before use in proprietary or restricted contexts.
Quickstart
pip install cua-agent
from cua_agent import Agent
agent = Agent()
# Configure with liteLLM model and run agentic workflow
Requires Python 3.11 or later (supports current versions up to <3.14). Depends on cua-core and litellm for core functionality.
Verify before relying
- Specific sandbox isolation guarantees and security model for computer interaction workflows
- Performance characteristics and latency for real-time agent operations across platforms
- Exact license terms and any commercial use restrictions
Package facts
| License | not declared (unclear) |
| Python support | supports the current Python release (<3.14,>=3.11) |
| Install friction | low — pure-Python wheel |
| Runtime dependencies | 12 — httpx, aiohttp, asyncio, anyio, typing-extensions, pydantic, rich, python-dotenv, cua-core, certifi, litellm, Pillow |
| Maintenance | actively maintained — 51 days since the last release |
| First released | |
| Downloads | 161,008/month — #10,648 on PyPI (30-day window, as of 2026-08-14) |
| Known vulnerabilities | none known (OSV.dev, checked 2026-08-14) |
Evidence: cua_agent-0.8.4-py3-none-any.whl
Tags
More Artificial Intelligence packages
LiteLLM provides a unified Python interface to…
permissive · top 100 on PyPI
huggingface-hubClient library and CLI tool for downloading,…
permissive · top 100 on PyPI
langchainLangChain provides a framework for building…
permissive · top 1,000 on PyPI
hf-xethf-xet provides chunk-based deduplication and…
permissive · top 1,000 on PyPI
tokenizersTokenizers converts raw text into token…
permissive · top 1,000 on PyPI
transformersTransformers provides a unified framework for…
permissive · top 1,000 on PyPI
cua-computerCua Computer provides a framework for…
unclear · top 15,000 on PyPI
cua-coreCua Core provides shared utilities, helper…
unclear · top 15,000 on PyPI
datarobot-genaiA toolkit for building and deploying AI agents…
permissive · top 15,000 on PyPI
llama-index-agent-openaiIntegrates OpenAI language models with…
permissive · top 5,000 on PyPI
strands-agentsStrands Agents is a Python SDK for building AI…
permissive · top 1,000 on PyPI
nvidia-natIntegrates enterprise agents with data sources…
permissive · top 15,000 on PyPI
agentscopeAgentScope is a production-ready framework for…
permissive · top 15,000 on PyPI
openhands-aciProvides tools and interfaces for AI agents to…
permissive · top 5,000 on PyPI
nova-actA Python SDK for building and running AI agents…
permissive · top 15,000 on PyPI
cugaCUGA is an enterprise-grade generalist agent…
permissive · top 15,000 on PyPI