From 90ab79445744688c90120563105efddf3880825c Mon Sep 17 00:00:00 2001 From: Leah Amelia Chen Date: Sun, 7 Dec 2025 12:04:50 +0800 Subject: [PATCH] CONTRIBUTING: tighten AI assistance disclosure requirements --- CONTRIBUTING.md | 23 ++++++++++++++++++++--- 1 file changed, 20 insertions(+), 3 deletions(-) diff --git a/CONTRIBUTING.md b/CONTRIBUTING.md index de7df4b71..0bc74af5d 100644 --- a/CONTRIBUTING.md +++ b/CONTRIBUTING.md @@ -23,10 +23,20 @@ it, please check out our ["Developing Ghostty"](HACKING.md) document as well. If you are using any kind of AI assistance while contributing to Ghostty, **this must be disclosed in the pull request**, along with the extent to which AI assistance was used (e.g. docs only vs. code generation). -If PR responses are being generated by an AI, disclose that as well. As a small exception, trivial tab-completion doesn't need to be disclosed, so long as it is limited to single keywords or short phrases. +The submitter must have also tested the pull request on all impacted +platforms, and it's **highly discouraged** to code for an unfamiliar platform +with AI assistance alone: if you only have a macOS machine, do **not** ask AI +to write the equivalent GTK code, and vice versa — someone else with more +expertise will eventually get to it and do it for you. + +Even though using AI to generate responses on a PR is allowed when properly +disclosed, **we do not encourage you to do so**. Often, the positive impact +of genuine, responsive human interaction more than makes up for any language +barrier. ❤️ + An example disclosure: > This PR was written primarily by Claude Code. @@ -36,6 +46,11 @@ Or a more detailed disclosure: > I consulted ChatGPT to understand the codebase but the solution > was fully authored manually by myself. +An example of a **problematic** disclosure (not having tested all platforms): + +> I used Amp to code both macOS and GTK UIs, but I have not yet tested +> the GTK UI as I don't have a Linux setup. + Failure to disclose this is first and foremost rude to the human operators on the other end of the pull request, but it also makes it difficult to determine how much scrutiny to apply to the contribution. @@ -45,10 +60,12 @@ work than any human. That isn't the world we live in today, and in most cases it's generating slop. I say this despite being a fan of and using them successfully myself (with heavy supervision)! -When using AI assistance, we expect contributors to understand the code +When using AI assistance, we expect a fairly high level of accountability +and responsibility from contributors, and expect them to understand the code that is produced and be able to answer critical questions about it. It isn't a maintainers job to review a PR so broken that it requires -significant rework to be acceptable. +significant rework to be acceptable, and we **reserve the right to close +these PRs without hesitation**. Please be respectful to maintainers and disclose AI assistance.