wuu documentation
Give an agent a task. Follow the files it reads, the commands it runs, and the changes it makes. Check the result in Review.


Wuu is an AI agent workspace for local projects. Give an agent a task, follow its file changes and commands, and return to the saved conversation when you want to continue. You choose the model service and the permissions for its work.
Start a task
Section titled “Start a task”The quick start takes you from installing the desktop app to checking your first result. The current release workflow produces an Apple silicon macOS preview; the app includes its own Wuu core. Terminal users can install the CLI from source.
For everyday work, learn how to select a workspace, manage conversations and forks, and inspect files, diffs, and command output.
Configure your workspace
Section titled “Configure your workspace”Model setup explains API connections, subscription credentials, and the difference between a provider and an agent engine. Read permissions and security before giving an agent access to private projects: local execution does not mean your model requests stay on the device.
Add reusable workflows with skills, connect external tools through MCP, or install plugins. For unattended scripts, use wuu exec. If a task fails, start with troubleshooting.
Develop with Wuu
Section titled “Develop with Wuu”The development guide covers building the app. Extension authors can start with an agent plugin or a desktop UI plugin; client authors can use the app-server protocol.