agentleFS
chanktbchanktb/claude-shopify-growth/skills/collection-content-deep/SKILL.md

collection-content-deep

Skill2 starsChanged 5 months agoMITView on GitHub

Generate long-form below-grid editorial content for Shopify collection mega-hubs (1,500–2,000 words across 8 sections: Category Overview, Brand Spotlight, How to Pick, Comparison Table, Application Guide, Care & Maintenance, Pricing Info, Related Blog Reads). Brand-specific knowledge prefill (fill in `references/brand-knowledge-base.md` with your own brands). Output HTML body suitable for Shopify metafield `custom.deep_content_html` rendered by the matching `collection-content` theme section. Pairs with collection-analyze and collection-audit-pipeline. Use when user says "write collection content", "deep content for collection", "long-form collection page", "expand collection beyond FAQ", "collection-content-deep".

The file

We split this skill into the rules it gives an agent

10 rules. About 0 are common, 6 are distinctive, and 4 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈6

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 4

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Very specific:

    Max Paragraph Length: Every paragraph <p> MUST be short (2-4 sentences max, or ≤80 words).

    line 190
  2. Very specific:

    Scannable Layout: Use bullet lists (<ul>/<li>), comparison tables, or bold lead-ins for key points to break up dense descriptions.

    line 192
  3. Distinctive:

    Ensure NO paragraph exceeds 4 sentences or 80 words.

    line 196
  4. Distinctive:

    No AI trigger phrases: "delve into", "in conclusion", "navigate the landscape", "in today's fast-paced", "it's important to note", "let's …

    line 197
  5. Distinctive:

    Check that content renders BELOW FAQ

    line 232
  6. Distinctive:

    Does NOT rewrite the tagline (that's collection.descriptionHtml — stays 20–40w per Cat 1.1)

    line 264
  7. Very specific:

    Does NOT generate FAQ (that's custom.faq_jsonld — separate metafield)

    line 266
  8. Distinctive:

    Does NOT modify product data

    line 267
  9. Very specific:

    Does NOT generate images (use blog-image skill for that — composable)

    line 268
  10. Distinctive:

    Does NOT pull GSC data (use Phase 4 GSC integration in collection-analyze)

    line 269

≈ “Common” is an estimate: at least 1 in 5 of the 364,980 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

Collection Content Deep: Long-Form Below-Grid Body Generator

Generates the 1,500–2,000-word editorial body that sits BELOW the FAQ accordion on a Shopify collection hub page. Solves the "thin content" problem that collection-analyze v0.2.1 left unaddressed — collections were getting a high composite score but only ~330 words of editorial content. This skill brings collection content depth on par with a well-written blog post while keeping the commercial-intent design (filter + grid still primary).

When this skill applies

Tier requirement: mega-hub or hub. Sub-tier collections (<100 products) do NOT need this depth — they get filtered by brand/season and serve as landing pages for specific intent, not authority hubs.

Tier rule check (per collection-analyze v0.2.1):

  • mega-hub (≥2000sp + OR-rule): MANDATORY apply
  • hub (1000–1999sp or generic category): RECOMMENDED apply
  • brand (100–999sp): OPTIONAL (apply for top-3 traffic brand collections only)
  • sub / tool / new: SKIP

Architecture

Content is stored in a collection metafield (custom.deep_content_html, type: multi_line_text_field or rich_text_field) and rendered by a theme section (sections/collection-content.liquid) placed AFTER the FAQ section in the template JSON.

templates/collection.<handle>.json
  order: [banner, product-grid, collection-faq, collection-content]
                                                       │
                                                       └─ reads collection.metafields.custom.deep_content_html

This pattern is consistent with how custom.sub_categories and custom.faq_jsonld work for the v0.2.1 hybrid template — metafield-driven, section-rendered, no template rebuild needed when content changes.

8 Sections (default template)

Each section gets a heading + body. Total target: 1,500–2,000 words.

Section 1: Category Overview (200–300w)

What is the category, why professional buyers + DIY shoppers use it, key benefits.

Field lesson: pricing-friendly + professional-buyer-focused. Address the buyer's professional decision, not just shopping inspiration.

Required elements:

  • 1-sentence definition of the category
  • 2–3 key benefits (commercial intent)
  • Quick-stat callout (e.g. "<N> in-stock SKUs across Brand A/B/C daily")

Section 2: Brand Spotlight (300–400w)

3–5 brands compared by personality. Each brand gets ~80w covering:

  • Founding / heritage 1 sentence
  • Positioning (premium / value / specialty)
  • Unique value (formulation / variant depth / price tier)

Knowledge base: see references/brand-knowledge-base.md — a template for you to fill in with your own brand profiles (Brand A, Brand B, Brand C, etc.).

Section 3: How to Pick (200–250w)

Decision tree by buyer use-case:

  • DIY / individual buyer — what to pick + why
  • Professional buyer (low volume) — what to pick + why
  • High-volume professional / business buyer — what to pick + why
  • Sensitive-use customer (allergy or sensitivity concerns) — what to pick + why
  • Beginner — what to pick + why

Format: <dl> definition list OR 3-column grid block.

Section 4: Comparison Table (150–200w + HTML table)

5–6 brands × 4–5 dimensions (performance metric, durability, variant range, price tier, sensitive-use / allergen-free status).

Table format: semantic <table> with proper <thead> / <tbody> / <th> scope. Styling inherits from theme.

Section 5: Application Guide (200–300w)

Step-by-step for the category's primary use-case (setup for a variant-rich product line, assembly for a kit, first-use steps for a technical product, etc.).

Format: ordered list <ol> with each step containing:

  • Step name (bold)
  • 1–2 sentences explanation
  • Optional pro-tip callout

Section 6: Care & Maintenance (150–200w)

How to extend product life / reset or refresh properly / store correctly. Reinforces post-purchase value, drives re-purchase signal to Shopify analytics.

Topics:

  • Longevity tips (usage habits, protective steps, etc.)
  • Maintenance best practice (cleaning/reset method per category)
  • Storage (temperature, light, shelf-life)

Section 7: Pricing Info (100–150w)

Your store's pricing model (retail, wholesale, and/or B2B) + volume discount tiers + free shipping threshold. Closes the commercial-intent loop.

Store-specific facts (fill in with your own policy):

  • Pricing model eligibility (open to everyone, or account-gated wholesale/B2B tier)
  • Free shipping threshold
  • Bulk / volume discount tiers
  • Dispatch time commitment (e.g. same-day cutoff)

Section 8: Related Blog Reads (50–100w + 3–5 link blocks)

Cross-link 3–5 of your own blog posts relevant to the hub. Increases internal PageRank flow to blog + improves content authority signal.

Format: <ul> with each link wrapped in <li> containing:

  • Article title (link)
  • 1-sentence teaser

Source: your blog article output directory (e.g. output/<site>/). Map by topic relevance.

Scoring Process

Step 1: Determine collection scope

Pull collection via Shopify connector:

  • handle, title, productsCount
  • existing descriptionHtml (tagline) — DO NOT overwrite, body content goes in metafield separate from tagline
  • existing metafield custom.sub_categories (sub-collection list informs comparison content)
  • existing metafield custom.faq_jsonld (FAQ themes already covered — don't duplicate in body)

Step 2: Identify brands present in hub

Scan first 50–100 products for vendor / brand tag distribution. Pick top 3–5 brands by product count → these are the brands featured in Section 2 + 4.

Step 3: Generate per-section content

For each of the 8 sections, generate text using:

  • Brand knowledge base (references/brand-knowledge-base.md)
  • Store-specific facts (pricing policy, shipping threshold, dispatch commitment, etc. — see Section 7)
  • Category-specific knowledge (e.g. the distinctions between product lines/variants/formats within your category, per the category-tier taxonomy)

Step 4: Em-dash + AI-trigger compliance & Readability Constraints (Paragraph Limits)

Rule: keep em-dash usage very low, ≤0.3 em-dash per 100 words across ALL copy. Use colons / periods instead.

Readability & Paragraph Rules (CRITICAL for mobile UX):

  • Max Paragraph Length: Every paragraph <p> MUST be short (2-4 sentences max, or ≤80 words).
  • Section Subdivision: Split each of the 8 sections (200-300 words) into 3-4 smaller paragraphs or block elements. Do NOT output a single wall-of-text paragraph.
  • Scannable Layout: Use bullet lists (<ul>/<li>), comparison tables, or bold lead-ins for key points to break up dense descriptions.

Self-check before push:

  • Count em-dashes in generated body (ratio ≤0.3/100w)
  • Ensure NO paragraph exceeds 4 sentences or 80 words.
  • No AI trigger phrases: "delve into", "in conclusion", "navigate the landscape", "in today's fast-paced", "it's important to note", "let's explore"

Step 5: HTML structure

Wrap each section in:

<section class="collection-content__section">
  <h2 class="collection-content__heading">{Section Title}</h2>
  <div class="collection-content__body">
    <!-- section content -->
  </div>
</section>

Theme section CSS handles styling.

Step 6: Push to metafield

mutation {
  metafieldsSet(metafields: [{
    ownerId: "gid://shopify/Collection/<id>",
    namespace: "custom",
    key: "deep_content_html",
    type: "multi_line_text_field",
    value: "<HTML body>"
  }]) { ... }
}

Step 7: Verify in preview

  • WebFetch the collection URL
  • Check that content renders BELOW FAQ
  • Validate semantic HTML (h2 hierarchy, table accessibility, no inline styles except for table layout)

Compose with other skills

  • Before: collection-analyze (identify which hubs need depth)
  • After: collection-analyze --re-audit (verify Cat 1.2 + Cat 4 lift)
  • Sibling: collection-refresh (Phase 2 — re-runs this skill for existing hubs with new brand data)
  • Reads from: brand-knowledge-base.md (per-brand profiles)
  • Writes to: metafield custom.deep_content_html per collection + optional theme template tweak (one-time)

Expected lift

Per collection-analyze v0.2.1 rubric:

CategoryBeforeAfterΔ
1.2 Below-grid copy4/4 (FAQ only)4/4 + bonus authority0 (capped)
2.9 Internal links3/33/3 (5 new blog links)0 (capped)
4.4 Schema bonus (HowTo)—+1+1
GEO/AI passage citability9/1010/10+1
Long-tail SEO surfacelowhighsignificant

The composite score won't change much (already high and capped). The long-tail keyword footprint and AI citation eligibility improve dramatically — that's the real value of deep content.

What this skill does NOT do

  • Does NOT rewrite the tagline (that's collection.descriptionHtml — stays 20–40w per Cat 1.1)
  • Does NOT generate FAQ (that's custom.faq_jsonld — separate metafield)
  • Does NOT modify product data
  • Does NOT generate images (use blog-image skill for that — composable)
  • Does NOT pull GSC data (use Phase 4 GSC integration in collection-analyze)

Output convention

Per collection artifact:

output/[site]/[handle]/
└── deep-content.md      ← markdown source (auditable)
└── deep-content.html    ← HTML body pushed to metafield

The markdown source is the canonical version; the HTML is generated from it. To update, edit the markdown + re-push the HTML.

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.