Document how AI is used here, and the review it goes through - #92
Merged
Merged
Conversation
This project uses an AI assistant for parts of its research, code and prose. docs/use-of-ai.md describes that: where it helps, what always needs a human, and the do's, don'ts and repo-specific tips, so a reader can weigh a protocol claim knowing the process behind it. Where a rule has an incident behind it, the page says so: an unmeasured claim presented as a finding, an empty-flight handshake_failure described as meaning one cause, a generated page rendered on one Python version, two similarly named discovery calls costing a reporter five weeks. A rule with no incident behind it is worth a second look. The page gives reasons alongside the instructions. A hedge is usually the symptom of a claim nobody measured, so a writing pass catches soft findings as well as soft prose. Shorter helps twice over: a reader hunting one fact finds it faster, and every restatement is a second place to update. Linked from a sentence in Contributing and as the last entry in the Documentation index, unbolded, so it stays findable without outranking the API reference. Prose is single logical lines, since the file is new and nothing here needs a fixed column.
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
This project uses an AI assistant for parts of its research, code and prose.
docs/use-of-ai.mddescribes that, so a reader can weigh a protocol claim knowing the process behind it: where the assistant helps, what always needs a human, and the do's, don'ts and repo-specific tips.Two things shaped the page.
Rules carry the incident that produced them. An unmeasured claim presented as a finding, an empty-flight
handshake_failuredescribed as meaning one cause, a generated page rendered on one Python version, two similarly named discovery calls costing a reporter five weeks. Each happened. A rule with no incident behind it is worth a second look.Reasons sit alongside the instructions. A hedge is usually the symptom of a claim nobody measured, so a writing pass catches soft findings as well as soft prose. Shorter helps twice over: a reader hunting one fact finds it faster, and every restatement is a second place to update.
Linked from a sentence in Contributing, where someone looks before opening a pull request, and as the last entry in the Documentation index, unbolded like the research notes so it stays findable without outranking the API reference.
798 tests pass on Python 3.11, 3.13 and 3.14;
check_share_safety.pyis clean; the link test covers both new references.