See below for our rules for interaction
Above all, SQLAlchemy places great emphasis on polite, thoughtful, and constructive communication between users and developers. Please see our current Code of Conduct at Code of Conduct.
The SQLAlchemy project takes a two-pronged approach to LLM use. We are fine if participants use LLMs on their own to understand new concepts, get answers to questions, or to help construct code for the purpose of issue demonstrations or even pull requests, however we must require that participants do not use LLMs as their primary means of communicating with humans on our github repositories or anywhere else.
This means:
-
no unsolicited LLM comments / answers on discussions or issues. When a human user posts a question on our discussion board, they've chosen to get a human answer to their question. If they wanted to talk to a chatbot, they can easily do that on their own! Users who run LLMs to auto-submit answers to discussion questions, "confirmed!" posts on bug reports, evaluations of bug reports, anything like that, will be banned, as fast as we see them. Again, you can use an LLM on your own to help with these tasks, but you must communicate with the github interface as a human, in a non-automated fashion.
-
no automatic PRs for open issues - for this one, we have an automatic close rule in place. See guidelines at Pull requests and AI contributions. Summary: we use LLMs here for simple changes already, and we have them customized to do things exactly as we want (not the least of which is that it's right here where I can interact with it immediately, rather than waiting a week for a contributor to come back to me who ends up being a robot anyway). You don't, so we don't really have much use for small PRs done by someone else's robot these days.