DOCS/contribute.md: set expectations for issues, PR text and LLM use - #18502
Conversation
|
While the AGENTS.md is well-intentioned, I see a few problems:
|
It's unavoidable anyway at this point.
You would hope so, but I don't know. And I know that
¯\(ツ)/¯ |
Drive-by contributions have become more and more problematic, so we need to make the rules more explicit. At the same time, we don't want to limit contributions, but make it clear that the submitter is supposed to take responsibility for the change they are sending. The biggest issue is overly verbose walls of text in issue or PR descriptions for trivial code changes. What's worse, a big part of the analysis is often simply wrong, making reviewers work twice as hard to untangle the mess. Contributions are welcome, but we have to be mindful of the time of everyone involved.
sfan5
left a comment
sfan5
left a comment
There was a problem hiding this comment.
lgtm, but explicitly explaining common sense is not going to stop slop cultists
My main expectation is that LLM actually will read that, and adjust itself a bit. |
3 participants
No description provided.