TAROTUI

A tarot reading experience in the terminal

Tip

TAROTUI in the master branch uses an ollama model with approximately 2-3gb of storage, that model is specifically and originally ollama3.2.
In the foreseeable future (yes i love using this word), options to choose custom models based on user preferences is going to be added to allocate different situations, such as not enough storage, etc.
Please know that qwen3.5:4b and deepseek-r1 support is added by switching to the qwen3.5 / deepseek-r1 branch and rebuilding the model through ollama create tarotui -f src/utils/Ollama_custom/Modelfile

• All currently supported models are listed below

Model	Size (in gb)	Status	Pros	Cons
DEEPSEEK-R1	5gb	Available at deepseek-r1 branch	Enhanced logic gate and detailed responses	Longer time for a POST request
llama3.2	2-3 gb	Available at master branch	Shorter time for a response, creative and quick	Less logic gate and reasoning and does not implicitly follow the Modelfile
qwen3.5:4b	3.4 gb	Available at qwen3.5 bramch	Detailed responses and improved logic gate in comparison to llama3.2	Extremely longer time for a POST request and is slower

For contributions in regards to model availability please read Contributing.md

• Sorry for the slow post request my internet is bunz :(

• TAROTUI is a tarot reading experience in the terminal built with python 🐍, questions are analyzed through ollama and card shuffling is done through the backend

Installation

Caution

For all universal linux distros, no specific distro package manager was used in the creation of this project, and thus requires python installed, This is a mere python program with no varying ux and incompatability of specific packages, making it available to systems with python installed

Using the shell script

• To begin your first tarot reading experience run the install script in shell using the following command, please note that this command does not account for git cloning, it assumes you've already git cloned the repository, if you havent git cloned the repository then git clone it now and follow up on the install script

• Please note that for the install, have python alongside pip installed to download the required libraries used to build this project

cd tarotui && chmod +x ./install.sh
./install.sh

If you havent git cloned then proceed to do so here and run the command above ^^

git clone https://github.com/Tsukeruu/tarotui

Manually

• For users who value security and do not trust shell scripts online, you may install it manually using the following set of commands
• First and foremost have python and pip installed, this section wont cover the installation for python as it requires covering for most distro package managers

cd tarotui
pip install -r libraries.txt

The command above installs all the python libraries required ^

if you have not installed ollama then proceed to do so using the following command

curl -fsSL https://ollama.com/install.sh | sh

After successfuly installing ollama proceed to build the tarotui model using the following command

cd tarotui
ollama create tarotui -f src/utils/Ollama_custom/Modelfile

The command above installs either ollama 3.2 / deepseek / qwen3.5 if not installed, which depends on your current branch, for further information head to the start of the README, disk space / usage may vary from model to model, and builds the custom model using the Modelfile which includes the persona required for the reading

Adding to PATH

• To add tarotui to path, create a shell script of the contents below
• Lets name this shell script... tarotui (Note you do not need to add a .sh extension due to the presence of the shabang)
• replace <INSTALL_DIRECTORY> with the absolute location of where you downloaded tarotui

Fish shell

#!/usr/bin/fish
python <INSTALL_DIRECTORY>/tarot.py

Bash shell

#!/usr/bin/bash
python <INSTALL_DIRECTORY>/tarot.py

Zsh shell

#!/usr/bin/zsh
python <INSTALL_DIRECTORY>/tarot.py

• After creating the shell script move it to a directory where it is visible in PATH such as ~/.local/bin, /usr/bin or /usr/local/bin
• And you should be good to go!

License

MIT