FAQ
Good to know before you begin
Permissions, local credentials and optional Android integrations.
Why does the app ask for permissions?
Most permissions are requested when a tool needs them. Media-read and microphone permissions are requested at startup. You can grant only the capabilities you use.
What does screen capture need?
Screenshots and recording use Android MediaProjection: grant the system prompt once, or authorize it later in Settings → Screen capture. Advanced Wi-Fi/Bluetooth switches are performed from the system settings pages.
Do I need Termux?
No for normal chat. The embedded userland is limited to arm64-v8a. External Termux RUN_COMMAND needs the Termux app and authorization. Package installation in the embedded environment may be unavailable on some builds; apt_status reports its state.
Where are my API keys stored?
In-app model settings and keys are saved on the device, not in bundled web assets. An optional config.json contains a plaintext key and overrides saved settings when read. Keep that file private and out of Git.
Is the app fully offline?
The frontend assets are bundled locally. Model requests use your configured endpoint, and features such as web search or map cards may need network access. A reachable local model endpoint can be used.
What about NFC writes and other changes?
NFC needs supported hardware; writes require confirmation. Tools carry a destructive marker for operations that should be confirmed, and the frontend marks write cards. Review proposed actions before approving them.
Can I install from Google Play?
The project distributes installable debug-signed APKs through GitHub Releases. The current target SDK and release signing are not suitable for Play Store distribution.