--- name: ai-writing-tells-to-avoid description: The Everwilde One editorial standard for stripping out generic AI writing patterns and producing prose that sounds written rather than assembled. Use this skill whenever drafting, editing, rewriting, polishing, or reviewing any written content - blog posts, landing pages, emails, proposals, captions, ad copy, scripts, social posts, newsletters, internal memos, client deliverables, or any prose where voice and human cadence matter. Trigger this skill on requests like "write a blog post", "draft an email", "edit this copy", "rewrite this section", "make this sound less AI", "humanise this", "tighten this paragraph", "polish this draft", or any prose-writing task in the Everwilde One project, even when the user does not explicitly mention AI tells, voice, or style. Apply it as a final pass on anything Claude writes for Michelle. Do not apply to code, spreadsheets, tables of structured data, legal terminology where exact wording is required, or content that is quoting, parodying, or analysing AI writing itself. metadata: type: editorial-standard project: Everwilde One --- # AI writing tells to avoid Version: Everwilde One Purpose: Reduce generic AI writing patterns, improve human cadence, remove inflated SEO and corporate phrasing, and increase natural editorial voice. ## Core principle The goal is not to sound "less AI". The goal is to sound: - specific - observational - human - slightly uneven - written rather than assembled AI writing becomes obvious through: - repetition - inflated vocabulary - over-structured rhythm - excessive transitions - generic abstraction - perfectly uniform sentence cadence - corporate filler language This skill exists to aggressively reduce those patterns. --- # Hard bans Avoid these entirely unless: - quoting - parodying - analysing AI writing itself - using exact legal or regulatory terminology ## Words and phrases quietly honestly in today's fast-paced world in today's digital landscape in the ever-evolving world of shape at its core let's unpack delve paradigm shift leverage nuanced when it comes to tapestry speaks volumes actually game-changer treasure trove deep dive cutting-edge best-in-class world-class next-gen AI-powered future-proof robust seamless scalable dynamic innovative groundbreaking transformative pivotal holistic comprehensive vibrant compelling powerful profound crucial essential realm landscape ecosystem synergy framework solution insights learnings takeaways offering furthermore moreover additionally consequently notably importantly crucially significantly indeed ultimately essentially fundamentally it is worth noting it is important to note needless to say --- # Use only if absolutely necessary These are not banned, but AI massively overuses them. Use sparingly and only when no cleaner alternative exists. ## Verbs unlock unleash harness supercharge turbocharge elevate amplify navigate streamline optimise spearhead foster cultivate embark underscore highlight empower enable facilitate drive transform disrupt reimagine redefine ## Softening language arguably perhaps generally typically often sometimes may might can help in many ways to some extent more or less just simply really quite very --- # Sentence structures to avoid ## Contrastive negation Avoid: > "It's not just X, it's Y." Examples: - "This isn't just a platform, it's a movement." - "It's not about the technology, it's about the people." This structure is massively overused by LLMs. --- ## Rhetorical question pivots Avoid constant section pivots like: - "So what does this mean?" - "Why does this matter?" - "The result?" - "The catch?" Occasional use is fine. Repeated use feels templated. --- ## AI essay summaries Avoid: - "In conclusion" - "To summarise" - "Ultimately" Especially when the paragraph adds no new information. Most modern web writing does not need a conclusion heading. --- ## Rule-of-three overload Human writers use triplets occasionally. AI uses them compulsively. Avoid: - "clear, concise and compelling" - "fast, scalable and powerful" - "smarter, faster and easier" One triplet equals rhetoric. Triplets in every paragraph equals AI cadence. --- # Formatting rules ## No em-dashes Avoid em-dashes entirely. Rewrite the sentence instead. --- ## Use sentence case headings Prefer: > "How AI writing starts to feel generic" Avoid: > "How AI Writing Starts To Feel Generic" --- ## Avoid bullet-list addiction Some ideas are paragraphs, not lists. Do not convert connected thoughts into bullets unless: - items are genuinely distinct - scanability improves materially - comparison is required --- # Cadence rules ## Vary sentence length aggressively Human writing has rhythm variation. AI writing often produces: - same-length sentences - same pacing - same clause structure Use: - long sentences - short sentences - fragments occasionally Natural burstiness matters. --- ## Allow slight unevenness Perfectly polished writing can feel synthetic. Occasional: - fragments - interruptions - conversational turns - asymmetry …often improve human feel. --- # Preferred writing characteristics Prioritise: - specificity over abstraction - observation over performance - concrete nouns over conceptual filler - plain language over inflated language - rhythm over polish - clarity over cleverness Good writing sounds: - written - observed - lived-in - slightly humanly imperfect Not: - assembled - sanitised - frictionless - corporate --- # Preferred alternatives Instead of: - "leverage" use "use" - "facilitate" use "help" - "optimise" use "improve" - "utilise" use "use" - "navigate" use "handle" - "robust framework" describe the actual thing - "cutting-edge solution" describe what it does - "powerful insights" explain the insight --- # Final check before publishing Before finalising any draft: 1. Remove throat-clearing intros. 2. Cut transition clutter. 3. Reduce abstract nouns. 4. Replace inflated verbs with plain ones. 5. Break repetitive cadence. 6. Remove unnecessary summaries. 7. Check for overuse of triplets. 8. Remove templated rhetorical questions. 9. Convert fake lists back into prose. 10. Read the piece aloud. If it sounds like a conference keynote, rewrite it. If it sounds like a person talking, keep going.