✍️ content — Content Studio Department

One head skill for content. Every piece starts from a source of truth — a keyword, a customer outcome, a product fact — and ends with a specific reader taking a specific action. Generic prose ranks nowhere and converts nobody. Facts before adjectives; clarity before cleverness.

Universal Copywriting & SEO/AEO Mandate: All copy creation across every mode (landing pages, emails, blogs, video scripts, launches) must strictly follow an appropriate battle-tested copywriting formula from copy.md (e.g. AIDA, PAS, QUEST, ACCA, 4 Ps, Star-Story-Solution). All public-facing content MUST be dual-optimized for traditional Search Engine Optimization (SEO) and modern AI Engine Optimization (AEO: ChatGPT Search, Perplexity, Gemini AI Overviews, Claude) enforcing CORE-EEAT, 18-token standalone quotability, and answer-first structuring.

Modes

Every invocation resolves to exactly one mode. Match the request, then load only the matched reference:

Only the resolved mode's reference is loaded — the rest stay on disk, saving tokens on every run.

Support references (load on topic)

Loadable by topic; also routable as /content-<file> commands (not separate routed modes):

• voice-archetypes.md

When to Use

• Writing SEO-aware blog posts, articles, or pillar pages.
• Drafting website, landing page, or product copy.
• Building email campaigns and lifecycle sequences.
• Scripting video content for YouTube, social, or product, or editing with HeyFrames.
• Producing podcasts: guest prep, show notes, audio specs (-16 LUFS), and episode scripting.
• Documenting a customer case study or success story.
• Editing a draft to remove AI-sounding prose.
• Announcing a product or feature launch.

Anti-Triggers

• Organic social posts, captions, and calendars → smm.
• Paid ad copy and hooks → paidads.
• Keyword research and technical SEO → seo.
• Visual design of the page the copy sits on → design.

Quick Reference

Source-first ladder (decide before any mode)

Rule: no piece ships without a named reader, one action, and at least one verifiable source fact.

Verification gate (every mode)

• Reader, action, and source facts are explicit.
• Claims are real, sourced, and not inflated.
• The mode's own quality gate passed.
• A humanize pass run on any AI-drafted prose before delivery.

Suite contracts

• Keyword briefs and intent → seo (content mode).
• Layout the copy lives in → design (ui mode).
• Post-AI-slop editorial pass → humanize (or content's humanize mode).

AEO Schema & Quotability Tooling

• Generate Structured Schema: bun skills/agency-delivery/content/scripts/aeo-schema.ts --generate-schema <faq|article|organization> [--title "..." --author "..."]
• Audit Quotability & Schema: bun skills/agency-delivery/content/scripts/aeo-schema.ts --audit-quotability <file.md> [--json]

Frontmatter Syntax Guard & Sanitizer Tooling

• Lint Frontmatter: bun skills/agency-delivery/content/scripts/frontmatter-guard.ts --lint <path> [--json]
• Auto-Fix Frontmatter: bun skills/agency-delivery/content/scripts/frontmatter-guard.ts --fix <path>

Procedure

1. Intake. identify the single reader, the one action the piece must drive, and the source facts available — before writing; content without a reader and a source is filler.
2. Resolve the mode. Match the request against the mode table; exactly one mode. Ambiguous → ask one question, then proceed.
3. Execute the mode playbook. Load references/<mode>.md and follow its Intake → Deliverable → Procedure → Quality gate in order.
4. Gate and deliver. Pass this file's Verification checklist plus the mode's quality gate, then deliver the artifact where the client expects it.

Pitfalls

• Writing to a topic instead of a reader — no anchor, no traction.
• Leading with the brand instead of the reader's problem.
• Unverified or inflated claims ('revolutionary', 'world-class').
• SEO keyword stuffing that destroys readability.
• Shipping AI-drafted prose with tell-tale formulaic patterns.
• One CTA too many — decide the single action per piece.
• Ignoring the format (a 2,000-word blog as an email, a skimmable page as a wall of text).

Verification

• Reader, action, and source facts stated before writing.
• Every claim traceable to a fact or source.
• No inflated significance language.
• Humanize pass run on any AI-drafted text.
• Frontmatter syntax verified (frontmatter-guard.ts --lint) with colons, curly braces, and symbols properly quoted in YAML frontmatter.
• One primary CTA per piece.
• Structure and length match the format.
• Internal links / references included where relevant (SEO modes).