Team playbooks

These playbooks answer one question per team: “What can I actually do with SEO Navigator MCP, and how do I ask Claude to do it?” Each page lists the features that matter to your role, the tools behind them, copy-paste prompts, and the traps to avoid — so a new teammate is productive in one sitting.

How to read a playbook

Every playbook follows the same shape: what you can do → prerequisites → core tools → example prompts → gotchas. Skim the prompts first; they show the fastest path to a result.

Pick your team

SEO team

On-page SEO (RankMath meta, schema, redirects, sitemaps), rank & traffic data (Search Console, GA4), topical maps and content briefs from QueryMind, and one-shot SEO reports. Open the SEO playbook →

Content team

Draft blog posts and pages (Gutenberg, Elementor, code-first), the GitHub markdown inbox shared with design, media and taxonomies, and QueryMind content jobs. Open the Content playbook →

CRM team

GoHighLevel contacts, opportunities and conversations, the weekly/monthly CRM-audit analytics, Google Drive audit docs, and Microsoft Clarity behaviour data. Open the CRM playbook →

Marketing team

Google Ads and Meta (Facebook / Instagram) Ads reporting, GA4 + Search Console, Clarity heatmaps, and funnel / opt-in / A2P campaign audits. Open the Marketing playbook →

IT / Admin team

The admin dashboard, sites / tokens / users, OAuth connections, per-site gated flags, the REST API, diagnose_site, and the security guardrails. Open the IT playbook →

The 60-second mental model

Everything below is shared across teams. Read it once.

1. One connector, many sites. You connect Claude to the SEO Navigator MCP server once (see Connect Claude). That single connection reaches every WordPress, Duda, Astro, and GoHighLevel site your token is scoped to — no per-site logins.

2. Sites are grouped into projects. A project is a client; a site is one web property or CRM sub-account under it. list_projects shows the client list with integration flags; list_sites shows the individual properties.

3. Everything is a “tool call”. You ask in plain language (“draft a post about X on the Acme blog”); Claude picks the right tool. You rarely type tool names — but each playbook lists them so you know what is possible.

4. Reads are free; writes are gated. Read tools (list / get / report) run immediately. Write tools ask for a confirm step first, and many writes are locked behind a per-site flag a super-admin must switch on (publishing, CRM writes, QueryMind jobs). This is intentional — the system is safe by default.

5. It never publishes-now or deletes on its own. Hard guardrails refuse publish-on-create and any delete of a post/page. Publishing is a separate, explicitly gated action. You can explore freely without risk of breaking a live site.

Prerequisites shared by every team

Before any playbook works you need: (1) Claude connected to the MCP server, (2) a token scoped to the sites you work on, and (3) the specific integration for your task configured on the site (e.g. a Google connection for GA4, a GoHighLevel PIT for CRM). The IT playbook covers all of these.

Where to go deeper

• All MCP tools — the full generated reference, with every argument and the risk tier of each tool.
• Error codes — what each E_* message means.
• Platform workflows — long-form, step-by-step guides per platform (WordPress, Duda, Astro, GoHighLevel).