What is TuriX-CUA
TuriX-CUA is an open-source AI intelligent agent based on Python that can achieve desktop automation operations through AI technology. TuriX-CUA "sees" the screen by taking screenshots, uses a multimodal large model to "think" about the next operation, and completes user-specified tasks, such as searching for information, processing documents, etc. TuriX-CUA supports Windows and MacOS systems, is compatible with a variety of AI models, and can integrate with other AI tools through the MCP protocol to automate complex tasks, providing users with an efficient and intelligent desktop operation experience.
Main features of TuriX-CUA
-
Screen recognition and operation: Recognizes screen content through screenshots, controls mouse clicks and keyboard input, and achieves automated tasks.
-
Multimodal task execution: Supports a variety of complex tasks, such as booking flight tickets, searching for information, generating charts, and inserting documents, etc.
-
Multi-model architecture: Uses Planner and Executor to divide work and cooperate, improving the success rate and stability of task execution.
-
Cross-platform support: Compatible with Windows and MacOS systems, adapting to the operating system needs of different users.
-
MCP protocol integration: Supports collaboration with other AI tools such as Claude through the MCP protocol, expanding functions and application scenarios.
-
Flexible model configuration: Users can replace the underlying AI model, such as OpenAI or locally deployed Qwen3-VL, to meet personalized needs.
-
Task planning and decomposition: Uses Planner to break down complex tasks into multiple steps and execute them one by one, improving task completion.
How to use TuriX-CUA
-
Environment preparation: Install Python and Conda, create a Python environment, and install the dependency packages required by the project.
-
Model configuration: Set the interface information of the AI model in the configuration file, and define the task description to be executed.
-
Permission settings: Grant the system permissions for screenshots and controlling the mouse and keyboard to ensure the software can normally operate desktop elements.
-
Start the Agent: Run the project startup script to let the AI begin executing the configured tasks.
-
Configure tasks: Modify the task description according to actual needs, ensuring the task statement is clear and explicit, and improving the success rate of execution.
TuriX-CUA project address
- GitHub repository：https://github.com/TurixAI/TuriX-CUA
Application scenarios of TuriX-CUA
-
Document processing: Automatically creates, edits, and sends documents (such as Word, Excel, PowerPoint), generates reports according to user instructions, inserts charts, and sends them to colleagues.
-
Email management: Automatically writes, sends, and replies to emails, and handles daily email tasks according to preset rules.
-
Data collection: Automatically extracts data from web pages or applications, such as searching and recording product prices, stock information, etc.
-
Content search and browsing: Automatically searches for content in the browser, browses web pages, and performs operations such as liking, commenting, or sharing.
-
Interface testing: Automatically tests the user interface of software to verify whether functions are running properly.