
Karpathy Coding Guidelines
Makes the agent write simpler, surgical code and verify its work against clear success criteria
- Repository
- sickn33/agentic-awesome-skills
- Stars
- 49,023
- Licence
- MIT
- Files
- 1 file
What it does
This skill gives the AI agent a set of behavioral rules for writing, reviewing, and refactoring code, based on observations by Andrej Karpathy about common LLM coding mistakes. Once installed, the agent is guided to state its assumptions and ask when unclear, avoid adding unrequested features or abstractions, change only the lines needed for a request, and turn tasks into verifiable goals such as writing a test first. It is a set of instructions only; it does not add new tools or automate anything by itself.
Great for
- Developers who want an AI assistant to make minimal, focused code changes
- Teams reviewing AI-generated code for overcomplication or scope creep
- Anyone refactoring existing code who wants behaviour preserved and checked by tests
Skip it if
- Quick throwaway prototypes or trivial tasks, where the guidelines add unnecessary caution (the skill itself says to use judgment)
- Projects that need specific architecture or style rules, since the skill is only general guardrails and does not replace those rules
Try asking
- “Add validation to this form and make sure the existing submit behaviour still works”
- “Fix the bug in this function by first writing a test that reproduces it”
- “Refactor this service with minimal changes and keep the existing tests passing”
- “Review this diff for overcomplicated code and unrequested abstractions”
Install in 3 steps
- 1
Download the skill
The ZIP contains the 1 original files, the MIT licence and an attribution note.
andrej-karpathy.zip - 2
Add it to your agent
Claude Code (all projects)
mkdir -p ~/.claude/skills && unzip andrej-karpathy.zip -d ~/.claude/skills/Claude Code / Agent SDK (this project only)
mkdir -p .claude/skills && unzip andrej-karpathy.zip -d .claude/skills/Claude apps (web/desktop): Settings → Capabilities → Skills → Upload skill, and choose the ZIP.
- 3
Ask for the task
No special command needed — the agent reads the skill's description and loads it when your request matches. Try one of the prompts above.
Get better results
- State the behaviour that must remain unchanged when asking for a refactor, so the agent has a clear success check
- Ask for a short numbered plan with verification steps for multi-step tasks
- Use it for real code changes; for exploratory or emergency work, say so so the agent can relax the caution
What's inside
- SKILL.md
Redistributed unmodified from sickn33/agentic-awesome-skills under the MIT licence. The assessment was drafted with AI from the skill's files on 2026-10-29. Skills can instruct your agent to run code — read SKILL.md and any scripts before installing.