FAQ

A local-first workspace for long projects that you work on together with an AI. Your project is an ordinary folder of Markdown and code on your disk. Git Lora builds a canonical model on top of it — entities, relationships, facts, events — so the AI answers from what you have already written instead of from whatever fitted into the last message.

A chat window rebuilds its understanding of your project every session, from whatever you paste in. Git Lora keeps that understanding as a versioned structure on your machine and hands the relevant parts to the model on demand.

The practical difference shows up in month three, when the project is too large to paste and you need the model to remember that a character died in chapter four or that a service was deprecated in March.

No. The same problem — a large body of work that an AI has to stay consistent with — appears in software, research, documentation and worldbuilding alike. Project kinds (Author, Coder, Universal) only choose which entity types are suggested at the start.

No. Branches inside Git Lora are a product feature, not Git commands, and nothing requires the terminal.

Because a project is a plain folder, you can run real Git over it if you want to. Git Lora does not hide your files in a private format, so both work together.

English, Russian and Simplified Chinese, in the desktop app and on this site. The language of your project content is a separate matter — that is whatever you write.