Stop AI Slop (Japanese)
Strips AI tells from Japanese prose: false agency, hedging, flat rhythm
Test report
- Verdict
- Tested · Works
- Score
- Tested
- Jul 17, 2026
- Environment
- Claude Code 2.x (agent harness)
- Upstream re-checked
- Aug 10, 2026 · 8ea564c
Task: rewrite a typical AI-slop Japanese post about リモートワーク (1470 bytes) — first with my usual 'make it more human' without the skill, then strictly following the skill's body; measured with a detector built from its own checklist and references/phrases.md. Original had 38 AI-臭 markers; baseline had 18 — it removed emojis, 定型句, and '3つのポイント', but left 7 偏愛語 (解像度/本質/手触り/熱量), propositional H2, and even increased 両論併記 from 4 to 5; the skill had 1 marker. Positive metrics, which the baseline completely lacked: 一人称 0→7, 伝聞 0→2, specific numbers 0→8; paragraph length variance (that very ムラ) 15.5 → 10.0 for baseline (became flatter = worse) → 38.6 for the skill, share of the most frequent ending 50%→42%→38%. On the skill's own scale, baseline scored 18/50, which was below its own threshold of 35 'rewrite from scratch', versus 42/50. I also checked the numerical claims in the docs: structures.md had exactly 6 系統, phrases.md 8 系統, examples.md 18 pairs against the promised 17 — it under-promised, not over-promised.
Scored on four weighted criteria — install, triggering, output vs. baseline, docs. How scoring works
- Installs cleanly 5/5
- Triggers reliably 5/5
- Output vs. baseline 9/10
- Docs & honesty 5/5
What Stop AI Slop (Japanese) does
Editing rules for Japanese text that go past em-dashes and pet words to the structural tells — absent author, thesis-style H2s, both-sides hedging, uniform rhythm. Triggers when drafting, editing, or reviewing Japanese writing that reads AI-generated.
How to install Stop AI Slop (Japanese)
git clone https://github.com/iKora128/stop-ai-slop-jp.git
mkdir -p ~/.claude/skills/stop-ai-slop-jp
cd stop-ai-slop-jp && cp -r SKILL.md references ~/.claude/skills/stop-ai-slop-jp/
Skills live in ~/.claude/skills/ (global) or .claude/skills/
(per-project). Restart Claude Code after installing.
Commands — how to trigger Stop AI Slop (Japanese)
-
/stop-ai-slop-jpStrips AI tells from Japanese prose: false agency, hedging, flat rhythm
It also activates on plain-language prompts like these:
-
このブログ記事、いかにもAIっぽさが抜けなくて硬いので、もっと自然に直してほしいのですが -
一人称も具体的な数字もない、ふわっとした文章をもっと人間らしく書き直してもらえますか -
段落の長さがどこも同じで単調で、AI臭いと編集長に言われた記事を直してほしいのですが
Frequently asked questions
- Is the Stop AI Slop (Japanese) skill free?
- Yes. The skill itself is free from iKora128/stop-ai-slop-jp. SkillProof publishes the install command and an independent test verdict at no cost.
- Does Stop AI Slop (Japanese) work with Claude Code?
- We tested it with Claude Code 2.x (agent harness) on Jul 17, 2026. Verdict: Tested · Works. Task: rewrite a typical AI-slop Japanese post about リモートワーク (1470 bytes) — first with my usual 'make it more human' without the skill, then strictly following the skill's body; measured with a detector built from its own checklist and references/phrases.md. Original had 38 AI-臭 markers; baseline had 18 — it removed emojis, 定型句, and '3つのポイント', but left 7 偏愛語 (解像度/本質/手触り/熱量), propositional H2, and even increased 両論併記 from 4 to 5; the skill had 1 marker. Positive metrics, which the baseline completely lacked: 一人称 0→7, 伝聞 0→2, specific numbers 0→8; paragraph length variance (that very ムラ) 15.5 → 10.0 for baseline (became flatter = worse) → 38.6 for the skill, share of the most frequent ending 50%→42%→38%. On the skill's own scale, baseline scored 18/50, which was below its own threshold of 35 'rewrite from scratch', versus 42/50. I also checked the numerical claims in the docs: structures.md had exactly 6 系統, phrases.md 8 系統, examples.md 18 pairs against the promised 17 — it under-promised, not over-promised.
- What is the Stop AI Slop (Japanese) SkillProof Score?
- 9.6/10 — installs cleanly 5/5, triggers reliably 5/5, output vs. baseline 9/10, docs & honesty 5/5.
- How do I install Stop AI Slop (Japanese)?
- Copy the install command from this page, run it in your terminal, and restart Claude Code. Skills live in ~/.claude/skills/ (global) or .claude/skills/ inside a project.
- Can I use Stop AI Slop (Japanese) with Cursor, Copilot, Gemini CLI, Codex or other AI tools?
- The SKILL.md format is native to Claude (Claude Code, Desktop, claude.ai). The instructions inside adapt to other assistants: Cursor rules, GitHub Copilot instructions, Windsurf rules, Custom GPTs, AGENTS.md for OpenAI Codex, and GEMINI.md for Google Gemini CLI — our conversion guides cover each, and the free converter on the tools page does the wrapping for you.