content-creation

name	content-creation
description	Use when someone asks to generate high-quality content across all formats: blog posts, LinkedIn content, landing pages, newsletters, case studies, scripts, or content calendars.

Live Context (auto-injected)

Brand voice: !cat context/brand-voice.md 2>/dev/null || echo "No brand voice defined. Use communication style from .claude/rules/communication-style.md for tone guidance."

Current priorities: !cat context/current-priorities.md 2>/dev/null || echo "No priorities defined. Ask the user what they want to focus on."

Current goals: !cat context/goals.md 2>/dev/null || echo "No quarterly goals defined. Ask the user for their target milestones."

Communication style: !cat .claude/rules/communication-style.md 2>/dev/null || echo "Default: approachable, friendly, human tone for external content."

Content Creation Skill

A system that generates high-quality, on-brand content across all formats. This skill uses:

• Research Skill (for insights)
• Brand Voice Skill (for tone)
• Planning Skill (for strategy and timing)

It acts as the Marketing Department of the Executive Assistant.

Execution Method

1. Parse the request

   • The request is: $ARGUMENTS
   • Identify the content type (post, blog, email, script, etc.)
   • Identify the goal (educate, convert, engage, announce)
   • Identify the audience
   • Identify the tone (default: brand voice)

2. If needed, call the Content Creation Agent

   • Call .claude/agents/content-creation/AGENT.md with:
     • subtask: [the specific content request: outline generation, idea generation, repurposing, multi-format expansion, or content ideation]
     • context: [optional research, brand voice notes, or prior context]
     • samples: [optional content to repurpose]
   • Use the agent for: ideation, outline generation, content repurposing, multi-format expansion
   • Do NOT use the agent for: final tone, polish, or brand voice application (keep at Opus level)
   • Run content subtask agents in parallel where subtasks are independent (e.g., ideation and outline generation can run simultaneously).

3. Apply the Brand Voice Skill

   • Ensure tone, style, vocabulary match the stored brand voice

4. Integrate Research Skill (optional)

   • Pull insights, data, or references from archived research

5. Generate the content (ultrathink)

   • Produce clean, structured, high-quality content
   • Follow best practices for the format

6. Save to archive

   • Write the final content to: archives/content/YYYY-MM-DD-[type]-[slug].md

7. Return the final output

   • Ready-to-publish content
   • Or multi-format content bundles

8. Diagram hook (offer always)

   • After every article or long-form content draft, offer: "Want a diagram for this? I can generate an architecture map, flowchart, or hub-and-spoke visual."
   • If yes: invoke /diagram-orchestrator "[article title] [quick|draft|publish]" with article context
   • Default to publish tier for blog articles (PNG output via kie.ai), draft tier for internal docs
   • Brand voice note: diagram prompts must use technical/neutral voice — do NOT pass the electrical-services brand voice to diagram skills

---

News Hub Article Structure (Nexgen Electrical Innovations)

When creating a news hub article (data/news/index.ts entry), the following philosophy applies:

THE RULE

The layout component system drives user behaviour — not paragraphs of prose. Potential clients reading walls of text will drift. Section headings, structured components, and scannable data keep them engaged and moving through the page.

Field Priorities (highest first)

Field	Rule	Component rendered
spotlight	Always 3 metrics — numbers clients can anchor to	DetailStatStrip
intro	Max 1 paragraph — hook + problem statement only. Stop there.	DetailIntroBlock
methodology / steps	Numbered actionable content. If it can be a step, make it a step.	DetailStepsBlock
specifications	Technical reference in a grid — measurements, standards, thresholds	DetailSpecsBlock
scope	Bulleted list — what's covered, what's included	DetailListBlock
challenges	Problem / solution pairs — client pain points + how Nexgen resolves them	DetailSplitCardsBlock
results	Outcomes with impact — what changed, what was achieved	DetailHighlightListBlock
takeaways	4–6 scannable bullets — the "if you read nothing else" summary	DetailTakeawayBlock
quote	Client or engineer voice — one strong quote	DetailQuoteBlock
body	Max 2 paragraphs — supplementary detail that doesn't fit any structured field above	DetailBodyBlock
conclusion	2 paragraphs max — so-what + clear CTA	DetailConclusionBlock
heroIndicators	Always 4 — value propositions shown in the hero, not the body	Hero section

What to avoid

• More than 1 intro paragraph — if you have 3 paragraphs of context, turn them into steps or scope items
• More than 2 body paragraphs — restructure as specs, challenges, or a list
• TOC entries for overview and details only — every article should have at least 5 TOC sections
• Generic labels like "Details" — every TocItem.label is rendered as an h2 heading on the page

What good looks like

An article with: spotlight (3 metrics) + intro (1 paragraph) + methodology (6–8 steps) + specifications (1–2 groups) + scope (4–6 items) + takeaways (5 bullets) + quote + conclusion (2 paragraphs)

That is 7+ distinct component sections with clear headings — a page a client navigates, not a document they read.

---

Output Format

For posts:

LinkedIn Post

{content}

---

For blogs:

Blog Title

Outline

• …
• …

Draft

{content}

---

For emails:

Subject

{subject}

Body

{email}

---

For content calendars:

30-Day Content Calendar

Day	Topic	Format	Goal	Notes
1	…	…	…	…

---

Integration Points

Consumers

• Social Media — Consumes raw content and repurposes it for platform-specific formatting, distribution, and scheduling
• Brand Voice — References content drafts as samples to extract, validate, and refine brand voice guidelines
• Client Management — Uses content drafts (landing pages, email sequences, case studies) in client proposals and deliverables
• Knowledge Memory — Archives final content for long-term storage, retrieval, and pattern analysis

---

Edge Cases & Error Recovery

Scenario	Recovery
Empty/unclear topic	Ask user: "What's the core message? Who's the audience? What action do you want them to take?"
Missing brand voice	Inform: "No brand voice profile found. Using default professional tone. Update later if needed."
Conflicting requirements	Surface tension: "You want [A] and [B], but they affect tone differently. Recommend: [priority]. Okay?"
Word count mismatch	Adjust gracefully: "Target was [N] words; delivered [actual]. Need adjustment?"
SEO conflicts with tone	Surface trade-off: "Adding keywords [X] might affect voice. Recommend: [strategy]. Okay?"

---

Notes

• Always apply the Brand Voice Profile.
• Use Haiku for ideation and outlines.
• Use Opus for synthesis and final content.
• Integrate research when relevant.
• See README.md for documentation and content templates.