code-review-pyramid

The Code Review Pyramid is a framework by Gunnar Morling that structures code reviews across five layers: API Semantics, Implementation Semantics, Documentation, Tests, and Code Style. Layers at the base deserve the most manual attention because they are the most expensive to change later.

Focus most on API semantics and implementation semantics, since design and logic flaws are hardest to fix after merge. Documentation and tests get moderate attention, while code style should be automated with linters and formatters rather than reviewed manually.

Check whether the API is as small as possible, offers one way to do each thing, follows the principle of least surprise, hides internals cleanly, avoids breaking changes to user-facing parts, and is generally useful rather than overly specific.

No. Code style sits at the apex of the pyramid and should be automated with formatters and linters wherever possible. Manual review effort should concentrate on the base layers like API and implementation semantics.

The pyramid guides general review prioritization but does not replace domain-specific checks such as security audits, compliance reviews, or performance benchmarking. Use it as a structuring framework alongside specialized review processes.