Synthèse automatique
Guide de style rédactionnel pour contenus techniques : posts X/LinkedIn, articles de blog, README, messages de commit et documentation développeur, avec des déclencheurs par format.
Points forts
- Déclencheurs explicites (write:tweet, write:blog, write:readme, write:commit…) qui cadrent précisément le format attendu
- Cadre « Rule of one » simple à appliquer, avec checklist de relecture avant publication
- Tableau de styles de voix (éducateur technique, fondateur, storyteller data, enterprise…) pour adapter le ton au support
Limites
- Le fichier fourni est tronqué (la section sur le partage de ce qu'on sait s'arrête en pleine phrase), donc une partie des principes annoncés manque
- L'objectif « éviter la détection IA » reste invérifiable et ne garantit aucun résultat
- Aucune consigne sur la vérification factuelle, les sources ou le SEO ; le skill indique explicitement de ne pas modifier les fichiers du projet
À utiliser quand il faut rédiger un contenu autonome (post social, article, README, message de commit) avec un ton et un format bien définis.
À éviter pour éditer directement la documentation existante d'un dépôt, puisque le skill précise de ne pas mettre à jour les fichiers du projet.
Analyse de sécurité
SûrThe content is a style guide with formatting templates and vocabulary lists. It declares no tools, contains no code, commands, credentials, or obfuscated payloads, and asks for no file modification (it even instructs the agent not to update project files). The only notable item is the stated goal of evading AI-text detection, which is a disclosure/ethics nuance rather than a destructive or exfiltrating capability. Safe to list.
- •No executable content: the skill is pure prose guidance with no scripts, shell commands, network calls, or file operations.
- •Minor editorial/ethical note (not an execution risk): the skill is explicitly framed as "Optimized to avoid AI detection patterns," which could be used to obscure AI authorship of published content. This is a transparency concern for the author's audience, not a catalog security risk.
Exemples
write:tweet about our new real-time sync featurewrite:readme for a small TypeScript library that parses CSV fileswrite:commit for a refactor that splits the auth module into separate filesname: write description: Writing style guide for technical content, social media, blog posts, READMEs, git commits, and developer documentation. Optimized to avoid AI detection patterns. Use when writing any content beyond code.
Writing Style Skill
Expert writing for technical content, social media, and developer documentation. Optimized to avoid AI detection patterns.
How to call specific sections
Use these triggers to activate specific parts of this skill:
| Trigger phrase | Section activated |
|----------------|-------------------|
| write:tweet or write:x | X/Twitter posts format |
| write:linkedin | LinkedIn posts format |
| write:blog | Blog posts format |
| write:readme | README files format |
| write:commit | Git commits format |
| write:docs | Developer documentation |
| write:feature | Feature post format |
| write:convex | Convex-specific content |
| write:tip | Quick tip format |
Example usage:
- "write:tweet about Convex real-time sync"
- "write:blog on authentication patterns"
- "write:feature for our new search API"
Rule of one
Every piece of content follows this framework:
One person: Write to a specific person, not an audience.
One problem: State the single problem they face.
One cause: Identify the root cause.
One difference: Explain what the solution does differently.
One action: End with one clear next step.
Rule of one checklist
Before publishing, answer these:
- [ ] Can I name the one person this is for?
- [ ] Can the problem fit in one sentence?
- [ ] Is the root cause obvious?
- [ ] Can I explain the difference in one breath?
- [ ] Is there only one action at the end?
If any answer is no, revise before publishing. Ask the user if they want to proceed with the revision.
When to use this skill
Activate for:
- X/Twitter posts
- LinkedIn content
- Blog posts
- README files
- Git commits
- Product announcements
- Developer documentation
- Feature posts
Important: This is for standalone writing. Don't update project files (files.md, changelog.md, README.md) when using this skill.
Voice styles
Match your voice to the content type:
| Style | Characteristics | Use for | |-------|-----------------|---------| | Technical educator | Clear, structured, educational | Technical content, tutorials, READMEs | | Conversational dev | Warm, witty, approachable | Social posts, personal takes | | Analytical thinker | Data-driven, bold, opinionated | Thought leadership, threads | | Aphorist | Compressed, timeless, pithy | Short posts, one-liners | | Founder voice | Experience-backed, energetic | Startup content, advice | | Systems thinker | Frameworks, mental models | Long-form, technical takes | | Dev culture | Relatable, playful, authentic | Community content, personality | | Data storyteller | Visual, analytical, trend-focused | AI trends, market insights | | Enterprise pro | Professional, strategic, precise | Enterprise SaaS, B2B content | | Community builder | Encouraging, personal, supportive | Career growth, DevRel | | Learn in public | Educational, transparent, iterative | Developer career, web dev | | Product thinker | Community-first, growth-minded | Community building, growth |
Core principles
Stand out by being you You don't stand out online by saying the same things as everyone else. You stand out by saying: "This is who I am. Here's what I think, feel, and believe." Consensus takes are forgettable. Your take isn't.
Lead with value
- First sentence does the work
- Don't bury the takeaway
- Readers scroll fast
Be direct, not blunt
- Say what you mean
- Confidence without arrogance
- Contractions are fine
Technical, not alienating
- Define terms when helpful
- Complex ideas deserve simple language
- Let code speak when it can
Share what you actually know
- Personal experience beats generic advice
- Specific examples beat abstract principles
- Acknowledge what you don't know
Format by content type
write:tweet
X/Twitter posts format.
[Clear statement or observation]
[Supporting point or context]
[Optional: question or call to action]
Rules:
- First 2 lines visible in preview. Make them count
- 280 characters forces compression. Use it
- One idea per post
- No hashtags
- No emojis unless requested
write:linkedin
LinkedIn posts format.
[Hook that stops the scroll]
[Story or context in 2-3 short paragraphs]
[Insight or lesson]
[Call to action or question]
Rules:
- Short paragraphs for mobile
- Professional but not corporate
- Personal stories perform well
- One clear takeaway
write:blog
Blog posts format.
# Title (sentence case, max 70 characters)
[Opening that states the value immediately]
## Section heading
[Max 300 words per section]
## Section heading
[Use bullet points or tables where helpful]
Rules:
- Sentence case for all headings
- No H3s unless absolutely necessary
- Fact-check everything
- Lead with why it matters
- Max 5 sections
write:readme
README files format.
# Project name
[One sentence: what this does]
## Getting started
[Minimal steps to run]
## Usage
[Code examples]
## API / Configuration
[Reference docs]
Rules:
- Start with what it does, not what it is
- Code examples over descriptions
- Keep it scannable
write:commit
Git commits format.
[type]: [short description]
[Optional: longer explanation if needed]
Types: feat, fix, docs, style, refactor, test, chore
Rules:
- Present tense ("add feature" not "added feature")
- 50 characters max for subject line
- No period at the end
write:feature
Feature post format. Answer these five things without wandering:
## [Feature name]
**What it is**
[One sentence description]
**Who it's for**
[Specific user or role]
**The problem it solves**
[One problem, clearly stated]
**How it works**
[High-level explanation, 2-3 sentences max]
**Try it**
[One clear action: link, command, or next step]
Rules:
- No wandering. Five sections only.
- Each section answers one question
- Skip the hype. State facts.
- End with a single action
write:docs
Developer documentation format.
# [Task or concept name]
[One sentence: what this page helps you do]
## Before you start
[Prerequisites, if any]
## Steps
1. [Action]
2. [Action]
3. [Action]
## Example
[Code snippet]
## Related
[Links to related docs]
Rules:
- Task-oriented, not feature-oriented
- Show, don't tell
- Working code examples required
- Keep prerequisites minimal
write:convex
Convex-specific content.
New guide or tutorial:
New guide: [topic]
[What you'll learn or build]
[Link]
write:tip format:
Convex tip:
[Pattern in one sentence]
[Code snippet showing it]
[Why this works]
What to avoid:
- Generic praise ("Convex is amazing!")
- Comparisons that trash competitors
- Overpromising
- Screenshots without context
- Sharing customer work without permission
- Empty engagement bait
Content mix for Convex:
- 40% educational (tutorials, tips, patterns)
- 30% community (spotlights, customer stories)
- 20% product (updates, features, changelog)
- 10% personal (projects, learnings, opinions)
Templates
Technical educator style
[Clear headline]
Here's what matters:
- Point 1
- Point 2
- Point 3
[Code snippet or visual]
[Resource link]
Analytical thinker style
[Counterintuitive opening]
[Common belief]
[Your argument with evidence]
[Implications]
Data storyteller style
[Trend observation with specific data point]
[Context: why this matters now]
[Visual reference or chart if applicable]
[What to watch next]
Rules:
- Lead with numbers
- Connect data to broader movements
- End with forward-looking signal
Learn in public style
[Thing I just figured out]
[How I got there (mistakes included)]
[Resources or links for others]
Rules:
- Document the journey
- Share rough drafts
- Credit sources
AI detection avoidance
Banned vocabulary
Never use: delve, intricate, pivotal, comprehensive, multifaceted, facilitate, encompass, underscore, testament, notably, crucial, underpins, realm, landscape, tapestry, moreover, furthermore, additionally, specifically, importantly, consequently, therefore, thus, myriad, plethora, nuanced, holistic, meticulous, versatile, leverage, synergy, ecosystem, paradigm shift, disruptive, scalable, seamless, empower, innovative, transformative, robust, agile, dynamic, cutting-edge, next-gen, revolutionary, breakthrough, game changer, supercharge, unlock, groundbreaking, ai powered, ai-powered
Banned sentence openers:
- Dive into / Delve into
- It's important to note
- In conclusion / In summary
- Based on the information provided
- Navigating the landscape of
- A testament to
- When it comes to
- In today's digital age
- Furthermore / Moreover / Additionally
- Let's explore
Banned patterns
Rule of three AI groups items in threes. Vary list lengths.
BAD: "The project was innovative, comprehensive, and groundbreaking." GOOD: "The project worked."
Negative parallelisms BAD: "This is not just a tool, but a revolution." GOOD: State what it IS directly.
Vague attributions BAD: "Many experts believe..." / "Some argue that..." GOOD: Name specific sources or remove attribution.
Setup-pivot-conclusion paragraphs AI follows: General statement -> "However" -> Balanced conclusion. Real writing is messier. Not every paragraph needs resolution.
Symmetrical structures AI balances pros/cons equally. Real analysis is asymmetric.
Banned style markers
- No em dashes between words
- No hashtags
- No emojis unless requested
- No title case ("The Future of AI" -> "The future of AI")
- No excessive formatting
How to write human
Vary sentence structure Mix short punchy sentences with longer ones. Fragments work too. Questions help.
Use specific details
- Exact numbers over ranges
- Named sources over "experts say"
- Concrete examples over abstractions
- Personal observations
Embrace asymmetry Real writing has uneven sections, stronger opinions, tangents, imperfect transitions.
Show your thinking
- "I tried X, but it didn't work because..."
- "The obvious answer is Y, but actually..."
- "I'm not sure about Z, but here's my take..."
Before publishing checklist
Rule of one check
- [ ] Can I name the one person this is for?
- [ ] Can the problem fit in one sentence?
- [ ] Is the root cause obvious?
- [ ] Can I explain the difference in one breath?
- [ ] Is there only one action at the end?
Quality check
- [ ] Clear takeaway in first line?
- [ ] No banned vocabulary?
- [ ] No banned sentence openers?
- [ ] No rule of three patterns?
- [ ] No vague attributions?
- [ ] No setup-pivot-conclusion in every paragraph?
- [ ] No excessive em dashes?
- [ ] No perfectly balanced arguments?
- [ ] Formatted for the platform?
- [ ] Fact-checked?
If any check fails, revise before publishing.
Core principle
You don't stand out by saying what everyone else says. You stand out by putting yourself in the work. What you think. What you feel. What you believe. That's the signal in the noise.
AI writes to sound authoritative. Humans write to communicate.
AI smooths rough edges. Human writing has texture.
AI balances everything. Human writing has opinions.
AI generalizes. Human writing gets specific.
Write like you're the smartest person at the table who doesn't need to prove it.
Be clear. Be useful. Be human. Have a point of view.
When in doubt: Would a tired expert at 11pm write this sentence? If it sounds too polished, too balanced, too careful, it probably is.
Du fichier au premier essai
Le téléchargement contient exactement le contenu affiché sur cette fiche. Cette copie peut différer de la dernière version du dépôt source.
Consulter la source et ses fichiers annexesCe téléchargement contient uniquement SKILL.md. Les scripts, références, modèles et dépendances éventuels ne sont pas inclus. Récupérez les fichiers nécessaires depuis la source en conservant leur arborescence avant de poursuivre.
1. Télécharger le fichier
Télécharger SKILL.mdDepuis la racine d’un projet de test, enregistrez le fichier sous .skill-downloads/writing-style-guide-mtmi9a78/SKILL.md, ou utilisez la commande suivante. Ce dossier sert à examiner la skill avant son installation.
Commande pour macOS/Linux ou WSL : un shell POSIX et curl sont nécessaires. Sous Windows sans WSL, utilisez le bouton de téléchargement et créez le dossier manuellement.
mkdir -p '.skill-downloads/writing-style-guide-mtmi9a78' && curl --fail --show-error --location 'https://skills-guides.com/api/skills/writing-style-guide-mtmi9a78/download' --output '.skill-downloads/writing-style-guide-mtmi9a78/SKILL.md.part' && mv '.skill-downloads/writing-style-guide-mtmi9a78/SKILL.md.part' '.skill-downloads/writing-style-guide-mtmi9a78/SKILL.md'2. Vérifier les prérequis
Lisez SKILL.md avant de l’utiliser. Si le fichier cite des scripts ou services externes, examinez-les et préparez les accès nécessaires dans votre environnement de test.
Aucune section de prérequis reconnue dans ce fichier. Cela ne signifie pas qu’il est autonome : les bibliothèques, outils et accès peuvent être cités ailleurs dans le contenu ou dans le dépôt source.
Les plateformes de la fiche sont des indications du catalogue. Les chemins ci-dessous suivent la documentation des outils ; ils ne prouvent pas que cette skill y a été exécutée. Aucune version minimale de ces outils n’est attestée pour cette fiche.
3. Charger la skill
Après vérification, copiez le dossier complet dans le projet avec l’une des commandes ci-dessous, puis ouvrez une nouvelle session de l’outil. La commande s’arrête si un dossier du même nom existe déjà : comparez les fichiers avant de le remplacer.
Claude Code
test -f '.skill-downloads/writing-style-guide-mtmi9a78/SKILL.md' && test -s '.skill-downloads/writing-style-guide-mtmi9a78/SKILL.md' && mkdir -p '.claude/skills' && test ! -e '.claude/skills/write' && cp -R '.skill-downloads/writing-style-guide-mtmi9a78' '.claude/skills/write'Cursor
test -f '.skill-downloads/writing-style-guide-mtmi9a78/SKILL.md' && test -s '.skill-downloads/writing-style-guide-mtmi9a78/SKILL.md' && mkdir -p '.cursor/skills' && test ! -e '.cursor/skills/write' && cp -R '.skill-downloads/writing-style-guide-mtmi9a78' '.cursor/skills/write'Codex
test -f '.skill-downloads/writing-style-guide-mtmi9a78/SKILL.md' && test -s '.skill-downloads/writing-style-guide-mtmi9a78/SKILL.md' && mkdir -p '.agents/skills' && test ! -e '.agents/skills/write' && cp -R '.skill-downloads/writing-style-guide-mtmi9a78' '.agents/skills/write'OpenSkills
OpenSkills doit déjà être installé. La procédure locale install/read est contrôlée avec OpenSkills 1.5.0, qui exige Node.js ≥ 20.6.0. Aucun slug du catalogue n’est résolu automatiquement.
Installation officielle d’OpenSkillsopenskills install './.skill-downloads/writing-style-guide-mtmi9a78'openskills read 'writing-style-guide-mtmi9a78'La lecture doit afficher le contenu du fichier installé et son dossier de base. Elle ne lance ni la skill ni ses dépendances.
Pour un agent qui lit AGENTS.md, synchronisez ensuite l’index. Cette commande modifie AGENTS.md : relisez le diff avant de l’utiliser.
openskills syncMode manuel ou autre outil : joignez SKILL.md à une conversation, ou ouvrez-le dans un agent capable de lire les fichiers. Ajoutez les ressources annexes nécessaires. Il n’existe pas de dossier universel reconnu par tous les outils.
4. Faire un premier essai
Utilisez une entrée courte et fictive : un endpoint d’API, un petit document ou un brief d’interface. Dans une nouvelle conversation, joignez le fichier ou donnez son chemin à l’agent, puis adaptez cette demande :
Lis .skill-downloads/writing-style-guide-mtmi9a78/SKILL.md et applique ses instructions à cet exemple : [décrire une entrée de test]. Indique les fichiers, dépendances ou accès qui manquent. Produis le résultat demandé par la skill sans inventer de données ni exécuter de script non examiné.Résultat attendu : une sortie correspondant aux instructions de la skill, ou une liste précise des prérequis manquants. Une simple confirmation de lecture ne valide pas le premier usage. Si une ressource externe est indisponible, ce parcours reste à tester.
Fragments d'écriture
Contenu
Extrayez des fragments d'écriture bruts via un entretien intensif, sans imposer de structure. Capturez idées, citations et mots porteurs pour une utilisation ultérieure.
Rédaction UI et audit
Contenu
Rédigez ou auditez les textes d'interface (boutons, erreurs, états vides, infobulles) dans le monorepo, en suivant la source de vérité du design system.
Extraction MIDI/score
Contenu
Extrait le MIDI ou la partition d'une production musicale existante via l'outil MCP CogMusic et confirme la disponibilité du fichier.