Mizchi Blog Style
Japanese style guide and AI-tone rubric for mizchi-authored blog posts (zenn / dev.to), calibrated on 18 articles / 937 sentences / 766 paragraphs. Scores a draft structure-first — paragraph granularity, heading form, code cadence, closing move — because marker rates (口語・文末・自分) can be hit exactly while the text still reads as pastiche. Use after drafting an article during the style-matching revision loop.
Install
npx skills add mizchi/skills --skill mizchi-blog-stylegit clone --depth 1 https://github.com/mizchi/skills.git /tmp/skills
mkdir -p ~/.claude/skills
cp -r /tmp/skills/mizchi-blog-style ~/.claude/skills/mizchi-blog-stylePersonal skills live in ~/.claude/skills. Use .claude/skills in a project folder to share a skill with your team.
From the SKILL.md
# mizchi Blog Style mizchi 個人ブログ向けの文体評価 skill。記事を書いた後に「mizchi 文体度」と「AI 臭度」を判定させて反復改善する。 一般的な日本語の自然文の規範は `natural-writing-ja`、AI 指数の測定手法は `ai-index` にある。 本 skill はその上に乗る**声固有の層**だけを扱う。 **すべての数値は実測**。zenn API から 18 本 (2025-06〜2026-09)、コードブロックを除いた 散文 937 文 / 766 段落 / 64,340 文字を計測した。再現手順は `../ai-index/references/calibration.md`。 ## 先に結論: マーカーを追うと失敗する この skill は当初、口語・文末・`自分` の出現レートで採点していた。それで書かせた記事が 「っぽい」と言われたので測り直したら、原因がはっきりした。 | 成分 | mizchi 実測 | 未編集 AI | マーカー最適化した原稿 | | --- | --- | --- | --- | | 文 / 段落 | **1.4** | 2.5 | **3.3** | | 一文段落の割合 | **64%** | 15% | 13% | | 5 文以上の段落 | **0%** | 0% | **23%** | | 冒頭段落の文数 | **1〜2** | 3 | 3 | マーカーレート (`自分` 11.1 対実測 11.5、文長 40 字対 43.6) はすべて目標圏に入れたうえで、 **段落の形は未編集 AI より遠かった**。粒度では AI のほうが正解に近い。 語句は合っていて形が違う文章は、パロディに見える。だから採点は構造から始める。 マーカーは**下限チェック**にしか使わない (§4)。 ## 1. 構造成分 (これが本体) `python3 ../ai-index/scripts/slopscore.py --lang ja draft.md` が全部出す。 ### 1.1 段落 = 1 文 **最重要。766 段落の実測分布:** ``` 1 文 = 64% 2 文 = 29% 3 文 = 6% 4 文 = 1% 5 文以上 = 0% 平均 1.45 文 / 段落 ``` 5 文以上の段落は 18 本で**ゼロ件**。3 文以上でも 7% しかない。 これは Zenn のレンダリング都合ではない。連続する `<p>` の元行番号を調べたら、 差が 1 (単一改行) のペアは **0%** で、全部に空行が入っていた。意図的な改行である。 書くときの操作はこれだけ。**文を書いたら空行を入れる。** 2 文つなげるのは、 2 文目が 1 文目の言い直しや直接の帰結であるときだけ。 **ただし「一文段落だから短い」は完全に間違い。** 313 個の一文段落を測ると: ``` 平均 48.8 字 中央値 43 70 字以上が 18% 十分位 [18, 27, 34, 38, 43, 51, 59, 68, 89] ``` 段落全体の平均文長 44.5 字より、一文段落のほうが**長い** (48.8 字)。 一文で切るからこそ、その一文が限定節を抱え込む。 長い一文は **が / ので / のだが / ですが** で従属節を中に持つ形になる。実物:
Details
- Skill name
mizchi-blog-style- Repository
- mizchi/skills
- Path
mizchi-blog-style/SKILL.md- Repo updated
- 18 November 2026
FAQ
What does the Mizchi Blog Style skill do?
Japanese style guide and AI-tone rubric for mizchi-authored blog posts (zenn / dev.to), calibrated on 18 articles / 937 sentences / 766 paragraphs. Scores a draft structure-first — paragraph granularity, heading form, code cadence, closing move — because marker rates (口語・文末・自分) can be hit exactly while the text still reads as pastiche. Use after drafting an article during the style-matching revision loop.
How do I install the Mizchi Blog Style skill?
skills CLI (Claude Code, Cursor, Codex and more): npx skills add mizchi/skills --skill mizchi-blog-style. Skills go in ~/.claude/skills (personal) or .claude/skills (project), and Claude loads them automatically when a task matches the description.
Is Mizchi Blog Style popular?
It lives in mizchi/skills, which has 423 GitHub stars and 6 forks.
Related skills
Actions CI Tuning
423mizchi/skills
Use when auditing or improving GitHub Actions workflows for a project. Covers cache setup (npm/pnpm/yarn), job parallelism, shard-based test splitting, artifact handling pitfalls, and Playwright-specific patterns. Trigger on: slow CI, cache miss, flaky shard jobs, merge-reports failures, or an explicit 'tune CI' request.
AI Index
423mizchi/skills
Method and tooling for measuring how AI-generated a piece of prose reads, in Japanese or English. Two layers — a deterministic lint for removable surface tells (scripts/slopscore.py, no API) and a judgment layer of typed probabilistic questions (Jev noul/score) for the tells that survive editing. Includes measured human/AI baselines, a gap-first calibration procedure, and labeled fixtures. Use when building or calibrating an AI-likeness score, or when a draft needs a defensible number rather than a vibe. For writing guidance use natural-writing-ja / natural-writing-en instead.
Apm Usage
423mizchi/skills
Reference for APM manifests, installation, targeted updates, deployment paths and lockfiles. Invoke when the user explicitly asks about APM or apm.yml, or an unfamiliar APM command/error needs exact CLI details. Do not automatically invoke for every project that happens to have installed skills.
Natural Writing En
423mizchi/skills
English prose norm for writing explanatory text that does not read as machine-generated. One diagnostic axis (does a sentence update the situation or the document?) plus measured thresholds for burstiness, hedging, enumeration and paragraph openings, with corrections to widely-cited signals that false-positive on human technical writing. Use when drafting or revising English articles, docs, READMEs or long-form explanations, especially after an LLM produced the draft. For Japanese use natural-writing-ja; for scoring a draft use ai-index.
Formal Methods Reconciler
423mizchi/skills
Use when reconciling software specs, docs, tests, configs, code, logs, or incidents with formal methods. Helps Codex extract claims, decide whether docs or implementation are the source of truth, choose an appropriate tool such as Z3, Alloy, TLA+, P, Dafny, MoonBit prove, Lean, Rocq, Why3, Verus, CBMC, Tamarin, or ProVerif, build the smallest useful model, run or plan verifier checks, and translate SAT/UNSAT, traces, proof failures, or proof obligations into domain-language questions and regression guards.
Natural Writing Ja
423mizchi/skills
Japanese prose norm for writing explanatory text that does not read as machine-generated. Built on one diagnostic axis (does a sentence update the situation or the document?) plus register purity, sentence-beat and 体言止め/反問 rules calibrated against 937 sentences of human technical writing. Use when drafting or revising Japanese articles, docs, or long-form explanations, especially after an LLM produced the draft. For English use natural-writing-en; for scoring a draft use ai-index.
Browse the most popular Claude skills, all writing & content skills, or write your own with the skills generator.