Commit Message Guide:

커밋하는 스타일을 개선해보기 위해 OpenAI 팀에서 사용한다는 커밋 스타일을 보고 학습하기위해 작성!

Commit Types

Type	Description
feat	Add a new feature
fix	Fix a bug
docs	Documentation changes only
style	Code style/formatting changes (no logic impact)
refactor	Refactor code without changing behavior
perf	Performance improvement
test	Add or update tests
build	Build system or dependency changes
ci	Continuous integration configuration changes
chore	Maintenance tasks (no code change)
revert	Revert a previous commit

These follow the Conventional Commits standard.

Format

<type>(<scope>): <short summary>

[optional body]

[optional footer]

Example

feat(api): add retry mechanism for failed requests

- Implements exponential backoff for network calls
- Adds configuration flag `maxRetryCount`

Examples from Real OpenAI-style Commits

Feature

1	feat(eval): add MMLU-lite benchmark for smaller models

→ Adds a new evaluation suite for lightweight models.

Fix

1	fix(api): correct error handling for streaming completions

→ Corrects bug in API streaming pipeline.

Refactor

1	refactor(trainer): simplify gradient accumulation logic

→ Cleans up internal training loop logic.

Performance

1	perf(model): optimize KV-cache reuse during decoding

→ Improves performance in GPT decoding.

Documentation

1	docs(readme): update setup instructions and directory structure

→ Updates documentation without code changes.

Test

1	test(tokenizer): add regression tests for unicode normalization

→ Adds new tests for tokenizer edge cases.

Build / CI

1 2	build: bump numpy dependency to 1.26.3 ci(workflow): enable scheduled benchmark runs

Maintenance

1	chore: remove legacy scripts and clean up project root

Revert

1	revert: feat(ui): add new onboarding flow

→ Reverts an unwanted feature addition.

Tips for Great Commit Messages

Use imperative tone (e.g., “Add”, “Fix”, “Refactor”)
Keep title ≤ 50 characters
Use blank line between title and body
Wrap body at 72 characters
Focus on WHY the change was made, not just what changed

Quick Reference Table

Type	Meaning
`feat`	Add new functionality
`fix`	Resolve a bug
`docs`	Documentation updates
`style`	Code formatting/style changes
`refactor`	Structural code improvement
`perf`	Performance optimization
`test`	Add/modify tests
`build`	Build or dependency changes
`ci`	CI/CD configuration updates
`chore`	Maintenance / cleanup
`revert`	Undo a previous commit

Summary

OpenAI and many large AI teams adopt this commit style for clarity, automation, and collaboration.
Using a consistent pattern like feat(animation): improve genie-lamp spawn transition helps generate changelogs, track intent, and maintain clean project history.