`/seo-site-structure`

Backed by `.claude/context/ai-seo/playbook/04-site-structure.md` (+ Topical Relevance prompt in `prompts/`).

Placeholders like ACME Agency, <id> and you@example.com mark values that are per-agency — your install fills them with YOUR clients and accounts. If a section references a helper script you don't have yet, it ships with that workflow's install.

/seo-site-structure — Core 30 + Topical Map

Backed by .claude/context/ai-seo/playbook/04-site-structure.md (+ Topical Relevance prompt in prompts/).

When to trigger

"Build the Core 30 for ACME Agency", "what pages should this site have", "make a topical map", "/seo-site-structure".
After /local-seo-audit flags a content-structure gap. Run this before /seo-content (structure first, then write the pages).

What it produces

A Google Doc "Site Structure (Core 30) — <Client>" with: the full page plan with target keyword + word-count per page, the URL hierarchy + internal-linking map, the topical map (topical clusters before geographic), location-page plan, and a gap analysis vs the current site.

Step-by-Step

1. Parse

First positional = client/business. Flags: --url, --location (primary city), --niche (e.g. "plumbing", "dental"), --dry-run.

2. Gather context

node ACME Agency/scripts/seo_run.mjs --skill structure --mode gather --client "<name>" --url "<url>" --location "<city>"

The gather scrapes the homepage and tries /sitemap.xml so you can audit existing pages. Read the CONTEXT.

3. Read the method

Read .claude/context/ai-seo/playbook/04-site-structure.md fully. Lock in the exact Core 30 composition (counts + word counts), the <id> rule, the internal-linking pattern, the GBP-proximity ceiling for location pages, and the sitemap rule. Load the verbatim Topical Relevance prompt via prompts/ and run it (mentally / with the niche) to generate the topical clusters.

4. Build the structure (markdown)

Page plan — table: | URL | Page type | Target keyword | Word count |. Category pages (3-10, 1,500-2,500 words), service pages (20-30, 1,000-2,000 words), GBP landing page, About, Contact, Locations.
URL hierarchy — the tree + the internal-linking pattern (which pages link to which; nav-link cap from field notes).
Topical map — clusters built for topical relevance first, derived from the Topical Relevance prompt for this niche.
Geographic layer — location/city pages, sequenced AFTER topical depth, each bounded by the GBP's real reach (don't plan city pages the GBP can't rank for).
Gap analysis — current pages (from the scrape/sitemap) vs the plan: keep / rewrite / create.

Write to ACME Agency/clients/<Client>/SEO/site-structure.md (or /tmp).

5. Save + report

node ACME Agency/scripts/seo_run.mjs --skill structure --mode save --client "<name>" \
  --in "<site-structure.md>" --title "Site Structure (Core 30) — <Client>" \
  --slack "*Site Structure (Core 30) — <Client>*
Plan: <N> pages (<c> category, <s> service). <k> already exist, <m> to create."

Add --dry-run to skip Drive/Slack. Output the Doc link.

Guardrails

Respect the ordering rule: topical relevance before geographic — never plan a geo layer with no topical foundation.
Proximity ceiling: don't promise/plan rankings in cities the GBP address can't reach.
Hand the page plan to /seo-content to actually write the pages.