Add policy on LLM contributions based on Zulip's

[jni]

As discussed in the core team Zulip channel.

We can iterate on this but we should get an MVP in soon (ideally in time for
0.7.0) and then discuss finer points of the policy and continue to make changes
as both we and the space evolve.

[TimMonko]

Thanks @jni! I'm glad we finally have the ball rolling on this. I think you are correct that we were trying to start by doing too much at once, and this creates a great foundation that we've agreed on.

For those unable to view the private Zulip chat, we discussed multiple different LLM policies from other large and small open source projects and agreed that Zulip captured our goal with an LLM policy the most.

[TimMonko]

Reading through differences from Zulips.
This is Zulip's:
Split up your changes into coherent commits, even if an LLM generates them all in one go.

I think that Juan's change is likely more fitting for our project, since we expect very little commit discipline

[TimMonko]

Suggested change

	write good code, including AI tools. However, as noted above, you
	always need to understand and explain the changes you're proposing to
	make, whether or not you used an LLM as part of your process to
	produce them. The answer to "Why is X an improvement?" should never be
	write good code, including AI tools. However, as noted above, you
	always need to make a sincere effort to understand and explain the changes you're proposing,
	whether or not you used an LLM as part of your process to
	produce them. The answer to "Why is X an improvement?" should never be

My only thought here is that I would have likely been a bit turned off by this. Not because I didn't understand the changes I was submitting, but because I didn't think it was possible for me to understanding everything without help. In other words, I was always afraid I was going to be wrong, and that made me hesistant to submit PRs, even if I did my very absolute best to understand things. (I do believe I was asking like chatgpt or something back a year ago when I was starting)

[jni]

Thanks Tim! I like the suggestion. (Will apply locally to have consistent line wrap.)

[psobolewskiPhD]

I would suggest going a step further:

Suggested change

	**Do not submit an AI-generated PR you haven't personally understood and
	tested**, as this wastes maintainers' time. PRs that appear to violate this
	guideline will be closed without review.
	1. **Do not allow AI agents to submit PRs for you.** Please submit any PRs yourself
	using the PR template.
	2. **Do not submit an AI-generated PR you haven't personally understood and
	tested.** Doing this wastes maintainers' time.
	PRs that appear to violate these guidelines will be closed without review.

[brisvag]

I like this! It forces that 1 extra reading step.

[psobolewskiPhD]

meta 😉

Suggested change

	documentation.
	documentation. When in doubt, please reach out on [our Zulip chat](https://napari.zulipchat.com).

[psobolewskiPhD]

Suggested change

	comments.
	comments.
	That said, feel free to use AI tools to proof read, correct, or improve your own English writing;
	we know it's not everyones native language and large language models (LLMs)
	can be very powerful and helpful in this domain.

[psobolewskiPhD]

Suggested change

	When you use an LLM to write a message for you, it remains **your
	If you use an LLM to write a message for you, it remains **your

[psobolewskiPhD]

I don't think we require screenshots do we?
I would consider removing this point, given my suggested changes in the warning box above.

[psobolewskiPhD]

I made some comments, mostly minor, but I do think it's worth -- at least for the time being? -- drawing the line on totally automated/AI PRs -- see also the related matplotlib matplotlib/matplotlib#31026

Putting that in the warning box and being able to explicitly reject based on that may make our lives easier.

[psobolewskiPhD]

the failed cis are because the new file isn't in toc:
https://github.com/napari/docs/actions/runs/21845649010/job/63040722497?pr=932#step:9:186

[jni]

@psobolewskiPhD Thanks for the review! I've incorporated nearly all of your suggestions, some with minor modifications. The exception is the language encouraging LLM use for English, which I'm at best ambivalent about. We can discuss it, but let's do so in a follow-up so this goes in ASAP!