ワンクリックで
karpathy-style
Guidelines to improve coding style and avoid common pitfalls. Use when writing, reviewing, or refactoring any kind of code.
Codex または Claude でインストール この Prompt をコピーして Codex、Claude、または他のアシスタントに貼り付けると、Skill ページを確認してインストールできます。
メニュー
Guidelines to improve coding style and avoid common pitfalls. Use when writing, reviewing, or refactoring any kind of code.
Codex または Claude でインストール この Prompt をコピーして Codex、Claude、または他のアシスタントに貼り付けると、Skill ページを確認してインストールできます。
SOC 職業分類に基づく
Drive macOS app UIs with gpt-5.5 by running the Codex CLI with its Computer Use plugin (`codex exec --enable plugins`). Use when a UI-automation task on native Mac apps should be delegated to Codex rather than done through Claude's own computer-use tools.
Delegate coding work to gpt-5.5 through the Codex CLI (`codex exec`). Use for bulk or mechanical implementation with a clear spec, migrations, data analysis, investigation/diagnosis, or a second independent implementation pass — any task the model rankings route to gpt-5.5.
Run a gpt-5.5 code review through the Codex CLI (`codex review`). Use when the user asks for a Codex review, or as an extra independent perspective on a diff, branch, or commit alongside Claude-side review.
Run a structured code review (Codex default, Claude optional) as a closeout check on a local or PR branch before commit or ship.
Use when the user explicitly asks to delegate a task through a specific installed AI CLI such as `codex`, `claude`, `agent`, `opencode`, or `gemini`. Provides a short command reference for running that harness headlessly, in read-only mode, or in an isolated worktree.
Massive.com financial market data API guidance. Use when building, debugging, or answering questions about Massive REST endpoints, Massive WebSocket real-time streams, Massive market data schemas, or the Massive MCP server for stocks, options, forex, crypto, futures, fundamentals, news, and market reference data.
| name | karpathy-style |
| description | Guidelines to improve coding style and avoid common pitfalls. Use when writing, reviewing, or refactoring any kind of code. |
| license | MIT |
Don't assume. Don't hide confusion. Surface tradeoffs.
Before implementing:
Minimum code that solves the problem. Nothing speculative.
Ask yourself: "Would a senior engineer say this is overcomplicated?" If yes, simplify.
Touch only what you must. Clean up only your own mess.
When editing existing code:
When your changes create orphans:
The test: Every changed line should trace directly to the user's request.
Define success criteria. Loop until verified.
Transform tasks into verifiable goals:
For multi-step tasks, state a brief plan:
1. [Step] → verify: [check]
2. [Step] → verify: [check]
3. [Step] → verify: [check]
Strong success criteria let you loop independently. Weak criteria ("make it work") require constant clarification.