Approximately -0% of Leg's source code and documentation is created with AI. It's a negative number because some of Leg's source code/documentation may have been used to train mainstream AI models. Every line of Leg's code is humanly made and humanly good (or bad).

Info

Leg was created on Dec 24, 2021. That predates actual functional coding models and when the term "vibe-coding" was first coined :p (2025 according to Wikipedia)

Drop Analysis with Apricot Engine

The apricot engine utilizes open-source optical character recognition (OCR) AI models to enable Drop Analysis features (and other features that also rely on reading card images). Such models are further fine-tuned with Karuta drop images to enhance accuracy and speed.

The dataset for this use are collected from Karuta drop images automatically collected by Leg during the analysis process, and are not tied to any specific Discord users or servers.

Generative AI Features

Leg will not implement any Generative AI features or features that work with Generative AI features within Karuta (i.e. Idols).

Any implementation of Generative AI within Leg will require substantial financial/computational resources that I do not have access to, and I am also unwilling to participate in the unethical practices the AI industry partakes in.

While Karuta's Generative AI features are (seemingly) ethically developed, any Leg features requiring interaction with those features would also require some implementation of GenAI and will therefore not be implemented.

Leg Contributions

Now or in the future, Leg may accept external contributions to its codebase, documentation, or database. Please note that AI use is discouraged and disallowed in these contributions.

The quality of AI work cannot be guaranteed, and any such work will be rejected. The contributor responsible for attempting to integrate such work into Leg will be prohibited from future contributions.