What problem does it solve? LLM-generated code often suffers from overcomplication, silent assumptions, scope creep, and unverified claims. This Skill provides a set of behavioral guidelines, derived from Andrej Karpathy's observations on LLM coding pitfalls, that keep AI-assisted coding surgical, simple, and verifiable. ## Core Features & Use Cases - Think Before Coding: Surfaces assumptions, ambiguities, and simpler alternatives explicitly instead of picking interpretations silently. - Simplicity and Surgical Changes: Enforces minimum-code solutions and restricts edits to lines that trace directly to the user's request, matching existing style. - Goal-Driven Verification: Transforms vague tasks into test-backed success criteria and requires adversarial repros for load-bearing safety or compatibility claims. - Use Case: While refactoring a validation module, the assistant writes a failing test first, touches only the relevant functions, removes only the imports its own changes orphaned, and builds a minimal repro to confirm the new input limit actually holds. ## Quick Start Ask the assistant to apply the karpathy guidelines while implementing or reviewing your next code change.
