IndigoCarmine (IndigoCarmine (Yuhei Yamada)) · GitHub
Skip to content
View IndigoCarmine's full-sized avatar

Organizations

@Project-HAK

Block or report IndigoCarmine

Block user

Prevent this user from interacting with your repositories and sending you notifications. Learn more about blocking users.

You must be logged in to block users.

Content in all repositories owned by your account will be closed.
Maximum 250 characters. Please don’t include any personal information such as legal names or email addresses. Markdown is supported. This note will only be visible to you.
Report abuse

Contact GitHub support about this user’s behavior. Learn more about reporting abuse.

Report abuse
IndigoCarmine/README.md

AIとCodingに対する宣誓

ここに置いてあるコードの一部はClaude Codeに書かせています。Commitを確認いただければ明らかでしょう。Agentic Codingの発展に伴い、人力のレビューは不可能になってきました。そこで、意図的にレビューしないことで開発の高速化を図っています。レビューするかしないかの基準はひとつだけで、バグが目に見える形で落ちるかで分けています。

GUIのレイアウト、CLIの出力整形、図の見た目、Blenderのスクリプト、ビルド周りは全部AIに任せていて、動いたらそれで終わりにしています。おかしければ実行した時点で気づくので。一方で、MDの計算設定の生成や解析、熱力学モデルのフィッティング、データの前処理は、生成させたあとで式のレベルまで自分で読んでいます。AIのPRはできるだけ厳しく審査していきます。ただ、全部読むやり方はコード量が増えると続かないので、検算できる形にしておくほうを重視しています。中間的な問題には、テストが期待の通りか確認しています。

以上は自分のリポジトリでの話です。ほかのOSSに関わる場合は、内容によらず確実に自分で確認していきます。

My policy on AI and code

Some of the code here was written by Claude Code, as the commit history makes clear. As agentic coding has advanced, reviewing everything by hand has become impossible, so I deliberately skip review in places in order to keep development fast. There is only one criterion for what gets reviewed: whether a bug would fail visibly.

GUI layout, CLI output formatting, how figures look, Blender scripts, and build tooling are left entirely to the AI, and once it runs I consider it done. Anything wrong shows up the moment I run it. Generating and analyzing MD simulation settings, fitting thermodynamic models, and preprocessing data are a different matter. I read those myself, down to the equations, after they are generated, and I review the AI's pull requests as strictly as I can. Reading everything doesn't scale as the code grows, though, so I put more weight on keeping the work in a form I can verify by independent calculation. For the cases in between, I check that the tests assert what I actually expect.

All of the above applies to my own repositories. When I contribute to someone else's open source project, I check the code myself without exception.

Pinned Loading

  1. chemical_structural_formula_viewer chemical_structural_formula_viewer Public

    Dart 2

  2. molecule_3d molecule_3d Public

    Dart 1