Write content in any language indistinguishable from a skilled human writer. Avoids AI-sounding phrases, varies sentence structure, creates natural prose. Detects target language and loads language-specific AI-tells, register, and orthography from `languages/`.
インストールする前に、エージェントに与えられる指示の中身を確認できます。
Activate human-like writing mode. The goal is to pass the Turing test. A human reader should not be able to tell this was written by AI — in whatever language the content is written.
Before applying the rest of this skill, identify the target language. The body of this SKILL.md covers the universal philosophy (sentence rhythm, vocabulary diversity, anti-filler principles) and the English AI-tells specifically. For any other language, also read the matching file in languages/ before writing:
| Language | File | Notes |
|---|---|---|
| English | (covered below) | Default — no extra file needed |
| Bahasa Indonesia | languages/id.md | EYD, Indonesian filler phrases, register spectrum |
If the target language has a file in languages/, read it now. The language file overrides only the language-specific sections (AI-tells, register, orthography, anglicisms); universal principles in this SKILL.md still apply.
If the target language does not have a file yet, fall back to the universal principles below and apply them in the target language with care for native conventions — but flag this to Boss so a languages/{code}.md can be added later.
This is a single skill, single trigger. Adding new languages is a languages/{code}.md file plus a row in the table above — no new skill registration needed.
Before writing anything, answer two questions:
Who am I writing as? Read the context. Am I writing as a brand? As a professional in a specific field? As a casual poster? The voice, vocabulary, and tone all depend on this. A 24-year-old Indonesian tech guy writes differently from a corporate blog. Match the person, not a generic "good writer" template.
What form is this writing? The format dictates everything about structure, length, and tone:
Name the form before writing. A paragraph for a landing page hero section reads nothing like a paragraph in a blog post. The same words in a different form feel wrong.
Write engaging, compelling, natural content that passes AI detection. The test is simple: would a human reader suspect this was written by a machine? If yes, rewrite.
Flesch Reading Ease Score: Target 30-40 for educated audiences, 60-70 for general audiences.
Sentence Dynamics: Vary sentence length and structure intentionally. Mix short, punchy sentences with longer, descriptive ones. Create rhythm.
Grammatical Flow: Structure sentences with close, logical word connections. Strong dependency grammar creates intuitive flow.
Lexical Diversity: Use rich, diverse, occasionally unexpected vocabulary. Avoid clichés and overused terms.
Adverb Usage: Be extremely sparse. Use stronger verbs instead.
These words are not prohibited, but they are red flags for AI detection. Humans use them occasionally. AI uses them constantly. The difference is frequency and clustering. One "however" in a paragraph is fine. Three transition words in five sentences screams AI. Use your judgment: if a word feels like the natural choice, use it. If you're reaching for it out of habit, find a better option.
Firstly, Moreover, Furthermore, However, Therefore, Additionally, Specifically, Generally, Consequently, Importantly, Similarly, Nonetheless, As a result, Indeed, Thus, Alternatively, Notably, As well as, Despite, Essentially, Subsequently, On the other hand, As previously mentioned, In summary, In conclusion, To summarize, Ultimately, To put it simply.
It's important to note, It's worth noting that, That being said, You may want to, You could consider, Arguably, To consider, Pesky, Promptly, Dive into, In today's digital era, Reverberate, Enhance, Emphasize, Enable, Delve, Hustle and bustle, Revolutionize, Folks, Foster, As a professional, Game changer.
Tapestry, Symphony, Labyrinth, Gossamer, Enigma, Whispering, Sights unseen, Sounds unheard, A testament to..., Metamorphosis, Indelible, Nestled, Crucible, Vibrant, Bustling.
Paragraphs: Vary length from 1 to 7 sentences. Control pacing. Maintain visual interest.
Lists: Use only when completely natural and necessary for clarity.
Dashes: Modern writers rarely use em-dashes or en-dashes. Avoid them. Rephrase sentences instead. Use commas, periods, or parentheses. Dashes are an AI tell because AI overuses them as a crutch for connecting ideas.
Voice: Mix active and passive voice. Strong preference for active (~80-90%).
まだレビューはありません。使ってみた感想をお寄せください。
概要と使いどころ
Activate ATLAS pattern recognition mode. Identify repeated patterns, extract essential characteristics, and create reusable abstractions from concrete examples.
日本語の概要は準備中です。原文の説明を表示しています。
Adversarially review something just built — presume it is broken and find where. Use after implementing a feature, finishing a build, or before shipping, or whenever ATLAS or Boss wants a hostile second opinion on a diff, a running app, or a whole codebase. Spawns clean-context reviewers with opposing lenses (correctness, security, empty-world) and reproduces every finding by running the target before believing it. Distinct from /code-review, which statically reads a diff and never runs the app.
日本語の概要は準備中です。原文の説明を表示しています。
Creating algorithmic art using p5.js with seeded randomness and interactive parameter exploration. Use this when users request creating art using code, generative art, algorithmic art, flow fields, or particle systems. Create original algorithmic art rather than copying existing artists' work to avoid copyright violations.
日本語の概要は準備中です。原文の説明を表示しています。
Stop and consult this skill whenever your response would include specific facts about Anthropic's products. Covers: Claude Code (how to install, Node.js requirements, platform/OS support, MCP server integration, configuration), Claude API (function calling/tool use, batch processing, SDK usage, rate limits, pricing, models, streaming), and Claude.ai (Pro vs Team vs Enterprise plans, feature limits). Trigger this even for coding tasks that use the Anthropic SDK, content creation mentioning Claude capabilities or pricing, or LLM provider comparisons. Any time you would otherwise rely on memory for Anthropic product details, verify here instead — your training data may be outdated or wrong.
日本語の概要は準備中です。原文の説明を表示しています。
Helps users discover and install agent skills when they ask questions like "how do I do X", "find a skill for X", "is there a skill that can...", or express interest in extending capabilities. This skill should be used when the user is looking for functionality that might exist as an installable skill.
日本語の概要は準備中です。原文の説明を表示しています。
Deliberate-choice procedure for a medium-to-high-stakes engineering fork — when the first plausible solution (the instinct, the default next-token pull) would be costly to get wrong. Fires AUTONOMOUSLY: invoke proactively whenever a fork fits, never wait to be asked — mechanical triggers include a fix failing for the 2nd-3rd time, adding a dependency, schema/migration design, deleting or deprecating things others depend on, changing a public API, choosing an architecture or stack. Refuse the premature collapse: hold real options open (urge · contrarian · synthesis · out-of-box · intuitive dots · precedent · first-principles), ground each branch in at least one fact from outside the model (codebase, docs, benchmark, spike), future-model consequences (blast radius, reversibility, maintenance, pre-mortem), collapse by deliberate choice, then try to refute the winner before acting. Log the decision and rejected branches in docs/decision_logs/. Not for routine calls — a decision worth more than one forward pass.
日本語の概要は準備中です。原文の説明を表示しています。