Package: opentars
Version: @VERSAO@
Section: utils
Priority: optional
Architecture: all
Depends: python3 (>= 3.8), python3-venv, python3-pip, python3-requests, python3-psutil, python3-pil, curl, ca-certificates, zstd, pciutils
Recommends: libxcb-cursor0, libxkbcommon-x11-0, libxcb-icccm4, libxcb-image0, libxcb-keysyms1, libxcb-render-util0, libxcb-xinerama0, libegl1, libgl1, python3-tk, python3-pil.imagetk, python3-xlib, python3-gi, gir1.2-atspi-2.0, at-spi2-core, xclip, gnome-screenshot | kde-spectacle | grim, wl-clipboard, libgtk-3-bin, libglib2.0-bin, libnotify-bin, xdg-utils, poppler-utils, alsa-utils | pulseaudio-utils
Suggests: ydotool
Conflicts: tars
Replaces: tars
Provides: tars
Installed-Size: @TAMANHO@
Maintainer: Enzo <enzorcasao@gmail.com>
Homepage: https://github.com/project-opentars/OpenTARS
Description: 100% local AI assistant for the Linux desktop
 openTARS is an AI assistant that runs entirely on your machine (through
 Ollama), with no cloud. It understands requests in Portuguese and English
 and controls the desktop: opens and closes
 apps, clicks buttons by name, types, searches the web, looks at the screen
 and runs terminal commands.
 .
 Its official AI is Gargantua (ollama pull ProjectOpenTARS/Gargantua), a
 small model trained to use openTARS tools, which the installer downloads.
 It also adapts to whatever AI models you already have in Ollama, picking
 the best one for each kind of task. Comes with a window ("opentars-gui"), a
 quick bar opened by a global shortcut ("opentars-gui --rapido") and a
 terminal mode ("opentars --chat") that also works on WSL and on Linux
 without a graphical session. Files (images, text, PDF...) can be attached
 to a request. The optional Cooper model (ollama pull ProjectOpenTARS/cooper)
 helps it look at the screen.