Works with the AI Agents you already use

    Claude CodeClaude CodeCursorCursorCodex CLICodex CLIDeepSeekDeepSeekMuseMuse+22 more

    Korean Copy Lint: Hype, Stock Phrases, Banned Words

    1

    They make a page read as machine-written and some of them are claims you cannot back up.

    Free

    Security scanned
    Korean Copy Lint: Hype, Stock Phrases, Banned Words

    Korean Copy Lint: Hype, Stock Phrases, Banned Words

    Example session with this skill installed

    배포 전에 랜딩 문구 검사해줘. "사장님"은 화면에 쓰지 않기로 했어. (Check the landing copy before deploy; we decided never to use the honorific 사장님 on screen.)

    • Read your context and instructions
    • Compiled the korean copy lint

    Summary

    Ran on a small test folder (a React hero, an HTML page, a markdown post). 1 fail, 9 warnings. Rewrites below; nothing changed until you pick.

    Project rule added

    copy-lint.json: { "banned": ["사장님"] }

    lint.mjs output (real run)

    △ Hero.tsx:3 [hype] 최초 — 업계 최초로 100% 만족을 보장합니다.
    △ Hero.tsx:3 [hype] 100% — 업계 최초로 100% 만족을 보장합니다.
    △ Hero.tsx:3 [hype] 보장 — 업계 최초로 100% 만족을 보장합니다.
    △ Hero.tsx:6 [stock] ~을 통해 — 다양한 기능을 통해 효율적인 업무를 경험하세요
    △ Hero.tsx:6 [stock] 효율적 — 다양한 기능을 통해 효율적인 업무를 경험하세요
    △ Hero.tsx:6 [stock] 다양한 — 다양한 기능을 통해 효율적인 업무를 경험하세요
    ✗ Hero.tsx:9 [banned] 사장님 — 사장님 전용 — 바로 시작
    △ Hero.tsx:9 [dash] — — 사장님 전용 — 바로 시작
    △ page.html:2 [hype] 압도적 — 압도적인 성능
    △ post.md:5 [hype] 획기적 — 이 글은 획기적인 방법을 소개합니다.
    
    파일 3개 · 실패 1 · 경고 9
    

    (✗ = fail, △ = warning, 파일 = files.)

    Not flagged, as intended: the comment // 최고의 주석..., width:100% in CSS, a <script> string, the front matter title, and "순위 보장은 하지 않습니다" (a stated limit).

    Proposed rewrites

    WhereOriginalRewrite
    Hero.tsx:3업계 최초로 100% 만족을 보장합니다.마음에 들지 않으면 첫 화면은 다시 만듭니다. (only if that is your real policy)
    Hero.tsx:6다양한 기능을 통해 효율적인 업무를 경험하세요예약 · 문자 · 정산을 한 화면에서
    Hero.tsx:9사장님 전용 — 바로 시작가게를 연 분을 위한 화면. 바로 시작
    page.html:2압도적인 성능첫 화면이 폰에서 1초 안에 뜹니다 (needs a measured number; tell me yours)
    post.md:5획기적인 방법(say what the method changes)

    Which rows should I apply? After editing I run the lint again until fails are 0.

    post.md

    Connects securely to your tools. The creator never sees your data.

    About this skill

    Korean marketing copy drifts toward the same few words: 최고 (best), 최초 (first), 100%, 보장 (guaranteed), and phrases that sound like a press release (다양한 기능을 통해 효율적인...). They make a page read as machine-written and some of them are claims you cannot back up. This skill finds them in the text readers actually see.

    What it does

    1. Reads only reader-facing text: string literals and JSX text in .js/.jsx/.ts/.tsx (comments skipped by a small scanner), text outside tags/script/style in .html, the body of .md/.mdx/.txt. Only pieces containing Hangul are checked.
    2. Five kinds of hits:
      • banned (fail): words from your copy-lint.json (old product names, honorifics you decided not to use, internal names that leaked to screens).
      • hype (warn): superlatives, guarantees, rankings. Skipped when the same sentence says what you do NOT do ("순위 보장은 하지 않습니다" is a limit, not a claim). 100% only counts next to words, so CSS widths are ignored.
      • stock (warn): empty template phrases.
      • dash (warn): em dashes in screen copy (turn off with allowDash).
      • long (warn): sentences over 80 characters.
    3. Reads each hit in context and sorts it: true and sourced (keep, add the source), unsourced (rewrite into what actually changes), stock (replace with what, how much, who).
    4. Proposes rewrites as a table (original -> rewrite). Applies only the ones you pick, then runs again until fails are 0.

    rules.md has the categories with before/after examples. Zero dependencies; add it to package.json scripts to run before deploy (fails only on banned words unless --strict).

    Rules it follows

    • Industry terms (platform, dashboard, API) are fine in the right place. The target is unverifiable claims, not vocabulary.
    • Never invents a number in a rewrite; asks you instead.

    The word lists are Korean. Instructions are in Korean.

    Source: https://github.com/Stophyuk/solutionzip-skills

    Find more skills like this on the Claude Skills Marketplace, the Codex Skills Marketplace and the Gemini Skills Marketplace, or browse Claude Skills for Prompt & Skill Engineering.

    Word lists target Korean text; English copy is not checked. The JS/TS scanner is a small tokenizer, not a full parser, so unusual syntax (regex literals with quotes) can confuse it. It flags words; deciding whether a claim is backed up is the agent's reading plus your answer. Instructions are in Korean.

    How to use

    1. 1

      Connect your agent to the Agensi MCP

      Copy the prompt and paste it into Claude, ChatGPT, Cursor or any agent that supports MCP. It sets up the connection for you. Manual setup

    2. 2

      Sign in to Agensi

      Your agent shows a sign-in link. Sign in and click Allow.

    3. 3

      Ask for what you need

      Your agent loads this skill and uses it on your task. It always asks before buying a skill.

    Reviews

    No reviews yet

    Be one of the first to try it. Every listed skill goes through the trust checks below.

    Security scanned

    Checked by our 8-point scan before listing

    Fresh listing

    Recently published to Agensi

    Free to use

    No account required to browse

    Trust & safety

    Security scan

    Security scan passed 16 days ago

    • Free with an Agensi account

    Software
    Node.js 18+

    Permissions

    Terminal / Shell
    Read Files
    Write Files
    Listed16 days ago

    What's inside

    Frequently Asked Questions

    No. It shows each hit with a proposed rewrite in a table and applies only the rows you pick. It never adds a number you did not give it.

    Yes. Put them in copy-lint.json at the project root: banned (fail), allow (remove a default word), allowDash, ignore (paths). Banned words fail the run, so it can sit in your deploy checks.

    This skill is compatible with any AI agent or CLI environment running Node.js 18 or higher. It is designed to work directly with local repositories, making it an ideal plugin for autonomous developers or localization agents.

    No, the skill is intentionally configured to ignore legal directories, Terms of Service, and Privacy Policies by default. This prevents the tool from accidentally flagging necessary legal jargon or contractual language that requires strict adherence to specific phrasing.

    You receive the full linting engine, a template for the `copy-lint.json` configuration, and lifetime access to version updates. The purchase includes the logic for superlative detection, filler removal, and structural analysis of Korean sentences.

    New in Prompt & Skill Engineering

    Free