The more fundamental question is: Is there information in the AI-coding session that should be preserved? Only if the answer is "yes", the next question becomes: Where do we store that data?

git is only one possible location.

I think there is very valuable information in session logs, like the prompts, or the usage statistics at the end of the session, which model was used etc. But git history or the commit messages should focus on the outcome of the work, not on the process itself. This is why the whole issue discussion before work in git starts is also typically kept separately in tickets. Not in git itself, but close to it.

There're platforms like tulpal.com which move the whole local agent-supported process to the server and therefore have much better after-the-fact observability in what happened.

I agree and like the analogy. And it's lossy to become useful as an AI in the first place. The "Learning" process has two effects on a machine learning model: In the beginning, it memorizes the facts it is trained on. But at some critical point, when it has no more capacity to memorize more facts, it starts to generalize. (This is why it's harder to train large models - large training datasets are needed.) And generalization is where AI models become very useful: For coding, writing poems, or any other task where memorization is not sufficient.

200k lines ... of which 85k lines are generated from Thrift and 10% from other projects, according to related ML discussions.

If you think the scale scales a bit strange at the upper scales, you are not alone.

Last year the scale went up to 300k, and someone didn't want to move all the lines.

Apache owns zero patents.

On the other hand, ASF is receiving major tax-reducable donations from US companies. The ASF is not in the patent business.

Google granted the license under the Apache License terms, and through distribution of Hadoop these extend to the end users of Hadoop under the very same Apache license. (AFAIUI, IANAL)