From karpathy-guidelines
Injects Lemon AI Hub's standard AGENTS.md and CLAUDE.md guidelines into the current workspace. Enforces Andrej Karpathy's surgical coding style, simplicity, and goal-driven execution.
How this skill is triggered — by the user, by Claude, or both
Slash command
/karpathy-guidelines:karpathy-guidelinesThe summary Claude sees in its skill listing — used to decide when to auto-load this skill
This skill injects the high-engineering standards of the Lemon AI Hub into your project. It ensures that any AI agent working on this codebase follows the **Andrej Karpathy surgical coding style**.
This skill injects the high-engineering standards of the Lemon AI Hub into your project. It ensures that any AI agent working on this codebase follows the Andrej Karpathy surgical coding style.
Use this skill when you want to:
Run the injection script from your terminal:
# From the root of your project
bash ~/.claude/skills/karpathy-guidelines/scripts/inject.sh
Note: The script will backup your existing
AGENTS.mdorCLAUDE.mdif they don't already contain the Lemon AI Hub marker.
Following Karpathy's recipe means:
npx claudepluginhub andersonlimahw/lemon-ai-hub --plugin karpathy-guidelinesGenerates or updates AGENTS.md/CLAUDE.md files via project scanning and interactive Q&A. Use when setting up AI-assisted development or updating project instructions.
Drop-in AGENTS.md that makes Claude Code, Codex, Cursor, and others behave like senior engineers — kills sycophancy, stops drive-by refactors, forces verification loops, and surfaces ambiguity. ~200 lines, works across many coding agents.
Crafts effective prompts, CLAUDE.md instructions, rules, agent system prompts, and skill bodies for reliable Claude Code behavior.