humanize-korean

humanize-korean is a skill for Claude Code from bam-bam-2/solo-skills. It costs 351 tokens per session (3,721 once invoked), scanned A, original, MIT.

A Korean-language editing workflow that makes AI-written Korean sound more natural while keeping its meaning unchanged. It checks for patterns such as translation-like wording, repetitive structure, excessive connectors, and overuse of emojis or lists.

In plain words
What is it for?
It is for polishing Korean text from tools such as ChatGPT, Claude, or Gemini, with a fast mode for shorter text and a stricter review mode for longer or more carefully checked text.
Why use it?
It helps remove writing habits that make Korean text feel machine-generated without changing the underlying content.

Skill for Claude Code

Written for Claude Code: ${CLAUDE_SKILL_DIR} variable. Also seen: reads .claude/ paths; mentions CLAUDE.md; mentions Claude Code.

Part of the solo-skills plugin — 26 skills shipped together

Good fit It is for polishing Korean text from tools such as ChatGPT, Claude, or Gemini, with a fast mode for shorter text and a stricter review mode for longer or more carefully checked text.

Compare 6 skills from other repositories ↓
Install with agentmods
npx agentmods add skills/bam-bam-2/solo-skills/humanize-korean
Install

Getting it into your agent

One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.

Any agent
npx skills add bam-bam-2/solo-skills --skill humanize-korean
Clone the repo
git clone --depth 1 https://github.com/bam-bam-2/solo-skills

Made for: Claude Code.

Or install solo-skills, the plugin that ships this one along with the rest of its 26 skills.

Wrote this? Show the measurements

A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.

agentmods badge for humanize-korean

README.md
[![agentmods](https://agentmods.dev/badge/skills/bam-bam-2/solo-skills/humanize-korean/github.svg)](https://agentmods.dev/skills/bam-bam-2/solo-skills/humanize-korean)
Your own site
<a href="https://agentmods.dev/skills/bam-bam-2/solo-skills/humanize-korean"><img src="https://agentmods.dev/badge/skills/bam-bam-2/solo-skills/humanize-korean/github.svg" alt="Measured on agentmods" height="20"></a>

Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.

agentmods 80×15 button for humanize-korean

Your own site · 80×15
<a href="https://agentmods.dev/skills/bam-bam-2/solo-skills/humanize-korean"><img src="https://agentmods.dev/badge/skills/bam-bam-2/solo-skills/humanize-korean.svg" alt="Reviewed on agentmods" width="80" height="20"></a>
Per session 351 Skills are progressive disclosure: only the name and description are preloaded; the body loads when the skill is used.
When invoked 3,721 The whole file, excluding the scripts and references it only reads on demand.
Security scan A 0 findings. A grade says what 26 rules found in the file — not that it is safe. Third-party audits
  • NVIDIA SkillSpector pass 7 Sept 2026
How audits are shown
Origin original No closer match found in the catalogue.
Token cost

What it costs to keep this loaded

Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.

ModelPer sessionOnce invoked
Fable 5.1 $0.00351 $0.03721
Opus 5 $0.00176 $0.01861
Sonnet 5 $0.00070 $0.00744
Haiku 4.5 $0.00035 $0.00372

Measured 12d ago against content hash 8a0d2a3cbee0, method: parsed. Prices are Anthropic first-party input rates as of 2026-09-11, from the pricing page.

Security

Grade A, and why

humanize-korean scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 12d ago.

The scan reads SKILL.md. This mod also ships 2 executable files (references/metrics_v2.py, references/metrics.py), listed below but not scanned — reading those needs a real analyzer, not pattern matching.

A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.

Nothing flagged

None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.

skills/humanize-korean/SKILL.md · 209 lines

How it starts

The opening of the file, as written. The whole thing — 209 lines — stays where its author put it; the contents beside it link to each section on GitHub.

Humanize Korean — AI 한글 티 제거 오케스트레이터 (v1.5)

v1.5 변경 고지 (2026-04-26) — v1.1 베이스라인 + Monolith Fast Path v1.2(voice profile)·v1.3(candidate pool)·v1.4(역할별 모델 분산)는 모두 핫패스 비용을 잡지 못해 5,000자 입력에 25분이 걸렸습니다. v1.5는 v1.1 단순 구조로 롤백한 뒤 단일 호출 monolith 에이전트만 추가한 설계입니다.

  • Fast 모드(디폴트)humanize-monolith 에이전트가 한 콜에서 탐지·윤문·자체검증 일괄 처리. 도구 호출 45회. 5,000자 이하 wall-clock 23분 목표.
  • Strict 모드(--strict) — v1.1 5인 파이프라인 그대로(detector·rewriter·auditor·reviewer + taxonomist 분류 자산 유지). 정밀 검증·장문(8,000자+) 처리·etc.
  • 삭제됨: voice profile·candidate pool·promotion-checklist·sample-collection·권한 위계 §1~§6.
  • 유지됨: 분류 체계 본진(C-9·C-10·D-7·H-3·I-3·I-4 등 v1.2~v1.3.1 신규 패턴)·rewriting-playbook·5인 에이전트 정의(strict 모드 백본).

Phase 0: 컨텍스트 확인 및 모드 결정

작업 시작 시 가장 먼저 다음 한 줄을 사용자에게 출력한다.

humanize-korean v1.5 — {fast|strict} 모드 / run_id: {YYYY-MM-DD-NNN}

모드 결정

  • 사용자가 --strict·"정밀 모드"·"5인 파이프라인" 명시 → strict
  • 입력 8,000자 초과 → strict (자동 승급 + 사용자에 1줄 고지)
  • 그 외 모두 → fast (디폴트)

run_id 결정

  • 모든 경로는 cwd 기준. 새 폴더 생성도 cwd 기준 _workspace/{YYYY-MM-DD-NNN}/에 만든다.
  • 기존 시퀀스 확인은 Glob 도구로 표지 파일을 매칭해 간접 조회. 올바른 사용법: Glob(pattern="_workspace/YYYY-MM-DD-*/01_input.txt") → 결과에서 폴더명 추출 후 NNN 최댓값 + 1. 주의: Glob은 디렉토리 자체는 매칭하지 못한다. 반드시 그 안의 표지 파일(01_input.txt)을 매칭할 것. Bash ls는 OS·셸 환경에 따라 경로 해석이 달라지므로 사용 금지.
  • 당일 폴더가 없으면 NNN = 001. 있으면 마지막 NNN + 1.
  • 부분 재실행 신호("이 카테고리만 다시"·"2차 윤문")일 경우 기존 run_id 재사용 + strict 모드로 자동 승급.

Fast 모드 (디폴트)

Phase 1: 입력 저장

  1. cwd 기준 _workspace/{run_id}/ 생성
  2. 입력 텍스트를 01_input.txt에 저장
  3. 첫 300자로 장르 자동 추정 (사용자 명시 시 우선)

Phase 2: Monolith 호출

humanize-monolith 에이전트를 Agent 도구로 1회 호출.

입력:

input_path: <abs path>/_workspace/{run_id}/01_input.txt
quick_rules_path: ${CLAUDE_SKILL_DIR}/references/quick-rules.md
genre_hint: 칼럼 | 리포트 | 블로그 | 공적 | null

출력 (에이전트가 직접 작성):

  • _workspace/{run_id}/final.md — 윤문본
  • _workspace/{run_id}/summary.md — 메트릭·자체검증·하이라이트

monolith는 단일 호출 안에서 다음을 모두 수행 (자세히는 에이전트 정의 참조):

  1. quick-rules 룰북 로드 → 메모리에서 패턴 탐지 + 윤문 + 자체검증 6항 점검
  2. 변경률 50% 초과 시 자동 롤백
  3. 자체검증 위반 시 1회 부분 재실행
  4. final.md + summary.md 작성

Read the full file on GitHub · 209 lines

Changes

What this file has done since we first saw it

Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.

  1. 12d ago First seen · 209 lines · 351 tokens per session scan A 8a0d2a3cbee0

Subscribe to this mod's changes

humanize-korean is a skill published in the GitHub repository bam-bam-2/solo-skills (362 stars, last pushed 8d ago), licensed MIT. It adds 351 tokens to every session and 3,721 once invoked, about $0.0018 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-30.