SEO Basics

What it does

SEO Basics audits or fixes the launch-level mechanics a public website needs before search engines and measurement tools can be trusted. It inventories the current owners first, expands real recipe dependencies, and verifies what the public site emits instead of treating an admin setting as proof.

A full run covers search access, metadata, tag management, consent, analytics, identity assets, base schema, and the WordPress-specific controls that apply. A named request runs only that concern and its dependencies.

Use it for

Use it before launch, after a redesign or migration, or when a neglected site needs a repeatable baseline:

• Audit the whole baseline without changing the site.
• Fix search access, metadata, sitemap, and robots alignment.
• Connect measurement and webmaster tools without duplicate properties or tags.
• Verify identity assets, schema, and applicable WordPress controls.

Recipe coverage

• Search access: indexability and canonical environment, sitemap and robots.txt, Google Search Console, and Bing Webmaster Tools.
• Measurement: Google Tag Manager, consent management, Google Analytics through GTM, and optional Microsoft Clarity through GTM.
• Identity and markup: core metadata, favicon family and manifest, social-sharing image, and truthful base schema.
• WordPress when applicable: Media Library alt text and the Rank Math baseline.

Missing access or an unapproved optional account becomes blocked or skipped. It never authorizes a fabricated ID, duplicate property, or secret in chat.

First run

Start with the public URL in analyze-only mode. The skill resolves the canonical environment, inventories the current owners, selects the requested recipes and their real dependencies, and records the baseline.

In authorized fix mode, it captures rollback evidence, makes changes in dependency order, verifies live output and provider state, then updates the durable report and machine evidence.

Inside the skill

The package separates the runtime router from individual concerns, authorization modes, shared standards, deterministic checks, and the contract that decides whether a run is actually finished.

What it creates

{project-root}/.seo-basics/
├── seo-basics.md
└── qa/
    └── live-validation.json

The Markdown file is the durable human report: selected recipes, dependencies, owners, findings, changes, verification, rollback, and open blocks. The QA folder holds machine evidence. Neither location may contain secrets.

Safety and boundary

Analyze-only mode can begin from the public URL. Google, Bing, CMS, DNS, and hosting access use provider-managed login/OAuth or the operating system’s protected credential store. Login, MFA, consent, DNS, and account-owner steps happen outside chat; no secret belongs in commands, Git, reports, or QA evidence.

Launch mechanics, not the whole SEO program.

Ready check

• Every selected recipe and dependency is PASS or evidence-backed N/A, with exactly one owner per output.
• GTM loads once and tested tag behavior agrees with the recorded consent decision.
• Metadata, assets, schema, sitemap, robots, and provider state pass live checks without duplicate output or IDs.
• Desktop and 375px renders are clean, and the report, source, platform state, and public response describe the same result.