Caveman (Claude Skill) — 설치 & 라이브 데모

왜 쓰나요

핵심 기능

Three intensity modes: Lite / Full / Ultra (+ Classical Chinese)
~65% average output-token savings (repo benchmarks)
Sub-skills: /caveman-commit, /caveman-review, /caveman-compress
Cross-platform: Claude Code, Cursor, Codex, Gemini, Windsurf, Cline, Copilot
Eval suite in-repo showing accuracy preserved, not just shorter

라이브 데모

실제 사용 모습

caveman-skill.replay ▶ 준비됨

0/0

설치

작성자 저장소 · Claude Skills 공식 문서

사용 사례

실전 활용법: Caveman

Get crisp one-line commit messages for every patch

👤 Anyone tired of 8-paragraph AI-written commits ⏱ ~5 min beginner

언제 쓸까: Claude writes a commit message, and it's a novel. You want it to say 'fix: off-by-one in pagination' and stop.

사전 조건

Caveman plugin installed — claude plugin marketplace add JuliusBrussee/caveman && claude plugin install caveman@caveman

흐름

Commit with caveman-commit

/caveman-commit — write a commit message for the staged changes.✓ 복사됨

→ One-line conventional commit; no body unless genuinely needed
Ship a branch of small commits

For the 4 staged files, use /caveman-commit per file — atomic commits, one line each.✓ 복사됨

→ git log shows 4 short commits

결과: Readable git log; bisect-friendly commits.

함정

Lost nuance when a message should be longer (e.g. breaking changes) — Override per-commit: 'write the full version, not caveman, this is a breaking change'

함께 쓰기: git · github

Review a PR with one-line actionable comments

👤 Reviewers (and reviewees who dislike AI walls of text) ⏱ ~15 min beginner

언제 쓸까: You're doing a quick pass and want pointed nits, not lectures.

흐름

Run caveman-review

/caveman-review — go through the diff. One line per comment.✓ 복사됨

→ Inline comments read like 'rename x → userId for consistency with api/users.ts'
Escalate when needed

On the auth file specifically, switch off caveman for a full-length security review.✓ 복사됨

→ Long-form only where the stakes justify it

결과: Faster review cycles; reviewees don't dismiss comments as noise.

함정

One-liners miss nuance on architectural changes — Architectural changes aren't Caveman's job — use default mode for those

함께 쓰기: github

Shrink CLAUDE.md and memory notes by 40–50% without losing intent

👤 Anyone whose CLAUDE.md grew past 5k tokens ⏱ ~10 min intermediate

언제 쓸까: Your project memory is eating context every session and you know it could be tighter.

흐름

Compress in place

/caveman-compress CLAUDE.md — keep rules and facts, cut the prose.✓ 복사됨

→ File shrinks ~46%; no rule deleted
Audit the diff

Show the diff. Any rule that changed meaning?✓ 복사됨

→ Per-change verdict; rollbacks if intent drifted
Commit

/caveman-commit — single-line commit for the compression✓ 복사됨

→ 'chore: compress CLAUDE.md (-46%)'

결과: Same rules, half the tokens.

함정

Compression drops a subtle caveat — Always diff-review after compressing rule files; intent loss is the main risk

함께 쓰기: filesystem

조합

다른 MCP와 조합해 10배 효율

caveman-skill + git

Terse commit workflow

Stage the change, then /caveman-commit, then git push.✓ 복사됨

caveman-skill + github

Caveman PR reviews at scale

For the 5 PRs in my review queue, /caveman-review each; post only actionable comments.✓ 복사됨

caveman-skill + claude-mem-skill

Shrink memory transcripts without losing decisions

/caveman-compress the claude-mem notes for this project; flag anything where meaning shifted.✓ 복사됨

도구

이 MCP가 노출하는 것

도구	입력	언제 호출	비용
/caveman	mode: Lite\|Full\|Ultra\|文言文	Toggle mode for the current task	free
/caveman-commit	staged diff (implicit)	At commit time; preserves staging	small model call
/caveman-review	PR diff	Quick review pass	small model call per file
/caveman-compress	file path	Shrink memory / rules files; diff-review before saving	small model call

비용 및 제한

운영 비용

API 쿼터: None
호출당 토큰: By design — negative; saves tokens
금액: Free (MIT)
팁: Use Ultra only on genuinely mechanical output (commits, tags); leave design discussions in default mode

보안

권한, 시크릿, 파급범위

자격 증명 저장: None

데이터 외부 송신: None — pure prompt shaping; no telemetry

Ultra mode can drop disclaimers and safety caveats. Don't use Ultra for legal, medical, or security advice output.

문제 해결

자주 발생하는 오류와 해결

Output still verbose after enabling Caveman

Most clients run Caveman per-turn; put /caveman Full at the top of each session or add it to CLAUDE.md

확인: First reply should feel distinctly terse

Caveman-commit returns multi-paragraph body

Usually triggers when the diff is genuinely large or breaks API. For true one-liner, add '—force-oneline'.

확인: Re-run with --force-oneline

Classical Chinese mode outputs garbled

Rare; usually a font / terminal issue, not the skill. Pipe to a file and open in an editor that supports CJK.

대안

Caveman 다른 것과 비교

대안	언제 쓰나	단점/장점
System prompt with 'be terse'	You want it without a plugin	Inconsistent; Caveman's intensity levels + benchmarks give measurable results
Custom slash commands	You only need one of the sub-behaviors (e.g. terse commits)	Reinventing the wheel

Caveman