First run

Select Add project, the folder button beside the project filter in the left sidebar, and choose a local folder. NativePi pins the folder so it is available the next time you open the app. You can also drop a folder anywhere in the window.

Pi asks for trust when a project contains local code or instructions, including extensions or skills. Trusting the folder allows those resources to load with your system permissions. Review the project before approving it.

Open Settings → Providers, choose a provider, and complete its authentication flow. Pi performs authentication and writes credentials to its own auth.json. NativePi does not copy credentials into renderer storage.

If you already use the Pi command line, your existing authentication is available immediately.

Select New chat. With several pinned projects, choose which folder to use; with one, the chat opens there. Choose a model and thinking level together from the picker beneath the input, then send a prompt. You can change the model or thinking level later, including while a turn is running.

Paste or drop an image into the composer to attach it. Drop another file to insert an @ path mention instead.

The sidebar groups chats into Pinned, Focus, and Finished. Mark a chat finished when you are done with it, or return it to Focus later. Right-click a chat to rename, pin, export, or manage it. The project filter changes the list; it does not move a chat or interrupt running work.

• Read NativePi and Pi to understand the ownership boundary.
• Read Sessions and storage before moving conversations between NativePi and the Pi CLI.
• Read Packages and extensions to add Pi capabilities.