Makes a public website mechanically search-ready through dependency-aware recipes, one owner per output, live verification, and a durable local report.
For: Site owners, developers, and SEO operatorsActs on: a public website and its search/measurement baseline14 recipes · 2 modes
The problem
Launch SEO often becomes a pile of unrelated settings: indexability conflicts, metadata has two owners, Analytics is installed twice, and the sitemap in robots.txt does not match the webmaster properties.
Saved admin settings are not proof that the public site is correct, and ad hoc fixes leave no durable record of dependencies, account ownership, consent, rollback, or verification.
This skill turns the baseline into 14 one-concern recipes, expands real dependencies, audits or fixes selected work, verifies the public result, and records one reproducible state.
Terminal skill
Runs in Claude Code, Codex, or Gemini CLI. Analyze-only can start from a public URL; fix mode uses the target's supported API, CLI, repository, or authorized admin surface. Secrets never enter chat or project files.
What it covers
The full baseline is broad enough for launch mechanics and deliberately stops before research, content strategy, or an advanced audit.
Search accessrequired
Indexability, canonical host, sitemap/robots, Search Console, and Bing.
Metadataone owner
Title, description, robots, canonical, Open Graph, and Twitter output.
Measurementthrough GTM
Tag Manager, consent, Google Analytics, and optional Clarity without duplicates.
Identity assetsapproved bytes
Favicon family, manifest, social image, and truthful base schema.
WordPress profileconditional
Rank Math, Media Library alt text, plugin checksums, and WordPress sitemap ownership.
Deep SEO strategyseparate work
No keyword research, broad crawl/CWV audit, content program, or advanced schema build.
Inside the skill
The package separates routing, one-concern recipes, authorization modes, shared standards, deterministic validators, and acceptance gates.
norml-seo-basics/
SKILL.mdruntime router, recipe catalog, and end-to-end baseline workflow
recipes/14 independent concerns with inputs, ownership, procedure, verification, rollback, and hard stops
processes/read-only versus authorized-fix behavior
standards/shared baseline contract and non-negotiable hard stops
scripts/dependency/catalog validator and live WordPress baseline validator
qa/ + templates/layered completion gate and reusable report/configuration files
Installed source stays separate from the per-site .seo-basics/ report and QA evidence.
How it flows
01
Resolve
Canonical target, platform, access, and project rules.
02
Inventory
Current public output and one owner per concern.
03
Select
Named recipes or full baseline plus dependencies.
04
Audit or fix
Read-only report or backup-first implementation.
05
Verify
HTTP, browser, providers, assets, schema, and parity.
06
Record
.seo-basics/seo-basics.md plus machine evidence.
Say this to Claude
“audit SEO basics for this public site
“run SEO basics and fix the safe failures
“connect Search Console and Bing without duplicate properties
“set up GTM, consent, Analytics, and Clarity correctly
“fix the sitemap and robots.txt mismatch
“review Rank Math and every WordPress image alt value
“verify the favicon, social image, and base schema
“reconcile the live site with the existing SEO basics report
What it does not do
Full technical audit — no broad crawl, Core Web Vitals, internal-link, or hreflang engagement.
Keyword/content strategy — no research, content plan, bulk metadata, or publishing workflow.
Invent business facts — no guessed identity, NAP, offers, reviews, schema claims, consent decisions, or provider IDs.
Bypass ownership or access — missing access blocks that recipe; it never justifies a duplicate property or secret in chat.