name: 9takes-editorial-standards description: The single source of truth for 9takes editorial rules - AI-writing tells to remove, banned words, voice attributes, frontmatter rules, and quality bars. Load when editing, polishing, de-AI-ing, or grading any 9takes blog content. Referenced by the editor agent and the editorial slash commands so the rules never drift between tools. path: .claude/skills/9takes-editorial-standards/SKILL.md
9takes Editorial Standards
One rulebook for every editorial tool (the editor agent, /deai, /copywriting-pass, /blog_content_editor_pass_people, /grade_blog, /cohesion-check). If a rule here conflicts with an inline rule in a command, THIS FILE WINS — and the command should be updated to match.
Hard rules (never violate)
- Never modify
lastmodfrontmatter. DJ manages it manually. No enrichment, editing, or automated pass touches it. - Em-dashes are banned in blog content. Zero per article (the blog lint enforces this). Replace with commas, periods, or restructure. This supersedes any older "1–2 is fine" guidance.
- Quality gate: overall grade ≥ 8.5 and discoverability ≥ 7 (per
/grade_blog) before a draft is publish-eligible. - Don't change the title or slug of high-traffic published posts (notably
enneagram-and-mental-illness) — light-touch edits only, SEO preservation first.
Brand voice
Tactically direct · socially savvy · respectfully provocative · pattern-recognition focused · results-driven. Writing rhythm: Hook → Insight → Action step. Key verbs: decode, navigate, map, read, unlock, resolve.
The three substance tests
Run the title, opening, headings, and every major claim through these:
- Can I visualize it? The reader should be able to close their eyes and see something. Abstract nouns (
growth,transformation,journey) fail. Zoom in until you land on a concrete object, person, scene, or behavior. - Can I falsify it? Claims should be verifiably true or false. "Helps you understand yourself better" fails; "Type 3s rehearse conversations before sending them" passes. Don't talk — point: a named behavior, a direct quote, a number, an observable action.
- Can nobody else say it? If Psychology Today or 16Personalities could publish the line unchanged, it fails. Find the 9takes angle: the anonymous Q&A mechanic, nine types answering one question, a specific type lens on a specific observable behavior.
AI-writing tells (kill on sight)
Sentence patterns
- Negative parallelism — "It's not X, it's Y" dramatic contrast (the #1 tell). Rewrite as a direct statement.
- [Statement] + [restatement in different words]
- Dangling -ing endings that fake depth: "...ensuring," "...highlighting," "...emphasizing," "...reflecting"
- False ranges: "From X to Y" implying a spectrum between unrelated things
- Weasel attribution: "Industry reports suggest," "Some critics argue," "Many experts believe" — name the source or cut
- Filler openers: "In today's world," "It's important to note/remember/consider," "At the end of the day," "When it comes to"
- AI transitions: "Let's explore/dive into/unpack/take a closer look"
- Compulsive summaries: reflexive "Overall," "In summary," "In conclusion" — conclusions add a new thought or CTA, never restate
- Essay-tone conjunctions: "Moreover," "Furthermore," "In addition," "On the other hand"
- Conversational remnants: "I hope this helps," "Certainly!," "let me know"
Word list (banned/flag): game-changer, deep dive, unpack, leverage, tapestry, landscape, navigate the complexities, at its core, shed light on, resonates with, multifaceted, delve, nuanced, realm, foster, pivotal, embark, myriad, paramount, groundbreaking, cutting-edge, intricate, underscore, enhance, crucial, testament (incl. "stands/serves as a testament"), captivate, solidify, cornerstone, spearhead, bolster, commendable, meticulous, watershed moment, enduring/lasting legacy, plays a vital/crucial role, leaves a lasting impact, rich cultural heritage/tapestry, vibrant community, breathtaking, must-see/must-visit.
Structural tells
- Lists of exactly 3 everywhere (triplets of adjectives, examples, benefits) — vary list lengths; sometimes 2, sometimes 4
- Every section the same length / same paragraph construction / same opening syntax
- The
**Bold Title:** definition. Elaboration.bullet template - Balanced "on one hand / on the other hand" everywhere
- Vague contextual opener paragraph that adds no information
- Title Case In Every Heading (use sentence case)
- Uniform energy throughout — vary intensity and sentence length (short punchy + longer flowing)
Formatting & readability
- Paragraphs 2–4 sentences; never multiple dense paragraphs back-to-back; mobile-first
- Headings tell a story when scanned alone; H2 = major section, H3 = sub-point; headings preview value ("Why Type 3s avoid vulnerability"), not labels ("Type 3 and vulnerability"); vary heading syntax
- Bullets for genuinely list-shaped content only — don't over-format prose
Repetition
Keep: thesis callbacks at structural moments, deliberate parallel structure, rhetorical emphasis with a new dimension. Cut: same point restated back-to-back with nothing new, identical phrasing recurring, multiple examples making one point without nuance, "in other words" + same idea. When cutting, consolidate into the single strongest version.
Frontmatter checklist
- title — works for humans (curiosity) AND search (key terms); specific, not generic
- description — the meta description: ~120–155 chars, core value proposition, makes someone click
- published — correct for the post's status (never flip without explicit approval)
- date — accurate; lastmod — DO NOT TOUCH
- enneagram — correct type number
Quality bar (post-edit)
Scannable in 30 seconds · flows read aloud · sounds like a human expert · value delivered by halfway · reader knows what to think or do next · each point made once, well.