소스 정보
- 저장소
- Dev-Toolbelt/dev-team-agents
- 최근 소스 활동
- 2026년 8월 20일 20:41
- 감지된 SKILL.md 언어
- 영어
- 스타
- 4
- 포크
- 0
설치 방법
기본적으로 소스를 먼저 확인하는 Prompt가 선택됩니다. 직접 명령으로 전환하거나 로컬 사본을 다운로드할 수도 있습니다.
소스 파일 검토
설치 여부를 결정하기 전에 SKILL.md와 SkillsMP에 표시된 보조 파일을 읽어 보세요.
메뉴
기본적으로 소스를 먼저 확인하는 Prompt가 선택됩니다. 직접 명령으로 전환하거나 로컬 사본을 다운로드할 수도 있습니다.
설치 여부를 결정하기 전에 SKILL.md와 SkillsMP에 표시된 보조 파일을 읽어 보세요.
Codex 또는 Claude로 설치 이 Prompt를 복사해 Codex, Claude 또는 다른 어시스턴트에 붙여 넣으면 Skill 페이지를 검토하고 설치를 진행할 수 있습니다.
직접 명령은 검토 Prompt를 거치지 않습니다. 실행하기 전에 소스를 확인하세요.
npx skills add https://github.com/Dev-Toolbelt/dev-team-agents --skill comments-policy명령은 한 줄로 유지됩니다. 복사하기 전에 가로로 스크롤해 전체 내용을 확인하세요.
로컬 사본을 원하시나요? SkillsMP에서 현재 제공할 수 있는 파일을 다운로드하세요.
Pure-markdown output rules for plans, reports, reviews, and any structured agent output.
Git worktree per task — <worktree_path>/<context>/<title>, in .worktrees.
Delegation mechanics — execution gate, autonomous sprint protocol, agent roster, spawn rules.
SKILL.md 표시 중
| name | comments-policy |
| description | Code comments — none by default; type annotations and AAA markers. |
The guiding principle: code should be self-documenting through clear naming. Comments are the exception, not the rule.
Code communicates intent through clear naming, small focused functions, and proper structure. Comments become outdated, double the maintenance burden, and often signal unclear code.
Before adding a comment, ask:
| Anti-pattern | Rule |
|---|---|
| Explaining what code does | Improve the code instead |
| Commenting bad code | Fix the code instead |
| Noise / obvious annotations | Remove them |
| Commented-out dead code | Delete it — Git history exists |
| Version-control comments | Use Git commits |
| TODO / FIXME | Create an issue tracker ticket |
Dividers (// === Section ===) | Use proper class/method structure |
| Journal / attribution | Use Git blame |
Closing brace markers (} // end if) | Refactor to smaller functions |
See references/anti-patterns.md for full before/after code examples.
| Case | Required? | Format |
|---|---|---|
| Business rule or regulatory constraint | Yes — explains WHY | Inline comment citing the rule |
| Non-obvious algorithm with external reference | Yes — explains WHY | Inline comment with URL |
| Workaround for a known external bug | Yes — explains WHY | Inline comment with ticket reference |
| Typed collections / generics | Yes — type annotation | Native language syntax |
| Exceptions / errors thrown | Yes — type annotation | Native language syntax |
| ORM/dynamic property types | Yes — type annotation | Native language syntax |
| Complex return shapes | Yes — type annotation | Native language syntax |
| Test structure | Yes — AAA markers | // Arrange, // Act, // Assert |
See references/type-annotations.md and references/aaa-pattern.md for full examples.
When a comment is warranted (WHY, not WHAT), it must be short: 1 line as the default, 2–3 only when the constraint genuinely cannot be stated in one. State the rule/workaround/reference and stop — no restating of the surrounding code, no walkthrough of alternatives considered, no multi-paragraph rationale.
Hard limit: max 3 lines per comment, max 3 comments per contiguous block (e.g. per config section, per function). A comment that needs more than 3 lines to justify itself belongs in an ADR (docs/development/adrs/) or a commit message, referenced by a single short line — not inlined. This applies everywhere, not only source code: config files (nginx, YAML, Dockerfiles), infra scripts, and templates follow the same limit.
| Anti-pattern | Fix |
|---|---|
| Multi-line header block above a config section explaining history/context | One line stating the constraint; move the rest to an ADR or commit message |
| Walking through what was tried before landing on this line | Delete — Git history already has this |
| Repeating the same WHY across several adjacent lines | State it once, near the first line it applies to |
Need to add a comment?
├─ Explaining WHAT code does? → ❌ Improve the code instead
├─ Explaining WHY (rule/workaround)? → ✅ Add it with context
├─ Type info the language can't express? → ✅ Required — use native syntax
├─ Exceptions / error conditions? → ✅ Required — use native syntax
└─ Test structure? → ✅ Required — use AAA markers
Detect the project's primary language and load the corresponding section:
| Language | Load |
|---|---|
| JavaScript / TypeScript | sections/javascript-typescript.md |
| Python | sections/python.md |
| Go | sections/go.md |
| PHP, Ruby, Java, C#, Rust, or other | sections/generic.md |
If multiple languages are present, load the section for the dominant one (most source files).
Load relevant sections based on context — don't load all sections eagerly:
| Context | Load |
|---|---|
| Python files in scope | skills/shared/comments-policy/sections/type-annotations.md |
Test files in scope (*.test.*, *_test.*, spec/) | skills/shared/comments-policy/sections/aaa-pattern.md |
| Legacy code review or refactor task | skills/shared/comments-policy/sections/anti-patterns.md |
| Greenfield development | None — core SKILL.md is sufficient |
references/aaa-pattern.md — AAA test pattern with examplesreferences/type-annotations.md — type annotation rules by languagereferences/anti-patterns.md — anti-pattern examples (what NOT to do)