We're glad you're here, and yes, that includes AI-assisted contributions. Used well, these tools are great. This document is about using them well.
Before you open your editor, tell us what you're planning. File an issue, comment on one that already exists, or kick off a discussion. A quick conversation up front lets a maintainer point you toward the right approach, flag work that's already underway, and save you from pouring hours into something that won't land. A large pull request that arrives with no prior context is hard to review and often has to start over, which is frustrating for everyone.
If an agent helped write your code, you're still the author. That means you can read the diff and explain why every line is there. Walk through the change, run it locally, and confirm it does what you expect. AI is a way to move faster on work you understand, not a shortcut around understanding it. When a reviewer asks "why this approach?", the answer should come from you.
The contributions that go smoothly tend to be small and self-contained. A bug fix with a clear cause. A doc that clears up a rough spot. A missing test. An accessibility improvement. Pick something you can take from start to finish and validate on your own machine. Sweeping changes that touch a dozen unrelated areas are hard to reason about and rarely merge as-is.
When you're discussing a PR or an issue, we want your thoughts — not a summary of what an AI thinks you should say. Even a short, uncertain reply moves things forward more than polished text you haven't read.
How a change was produced doesn't change what we expect of it. Every contribution is held to the same standard for clarity, scope, and fit with the project's conventions. Pull requests that show the author doesn't understand the change, that reach further than they need to, or that ignore how Kibana does things will be closed. That's not about the tools you used. It's about the result.
Thanks for pitching in. We're happy to have you.