karpathy-coder

Review staged code changes for unnecessary complexity by using Python-based analysis to detect premature abstractions, bloated designs, and diff noise. This process surfaces hidden assumptions and enforces surgical edits to keep your commit changes minimal and focused.

Diff hygiene is the practice of keeping code changes surgical by detecting unrelated edits and diff noise within your staged changes. Reviewing your diffs ensures that only necessary, focused modifications are committed, preventing hidden assumptions and unexpected side effects from entering the codebase.

Create a testable implementation plan by stating verifiable goals and clear success criteria before writing code. Reviewing your plan surfaces hidden assumptions and tradeoffs early, ensuring your multi-step engineering tasks remain disciplined and verifiable before execution.

Yes, you can check implementation plans for hidden assumptions before coding by applying reference guidance to evaluate your design. This surfaces tradeoffs and weak verification early, ensuring your multi-step engineering tasks have testable plans and clear success criteria before you start.

Yes, Python is required because the complexity analysis and reference guidance use Python-based scripts to evaluate your code changes. These scripts analyze your staged diffs and implementation plans to surface assumptions, simplify designs, and enforce surgical edits.

Code reviews miss weak verification when implementation plans lack clear success criteria and testable goals. Surfacing hidden assumptions and requiring verifiable outcomes before execution ensures your multi-step engineering tasks are fully validated and ready for commit.