Setup

Three pieces: the commands (slash commands for Claude Code), two MCP servers (live data: Google Search Console + DataForSEO), and the knowledge base (the methodology the commands cite).

1. Install the commands

Copy everything in commands/ to your Claude Code commands folder:

# global (all projects)
cp commands/*.md ~/.claude/commands/

# or per-project
cp commands/*.md your-project/.claude/commands/

That's it — they show up as /seo-analyze, /seo-quick-wins, etc.

2. Google Search Console MCP (free, required)

The commands call tools named mcp__gsc__*, so the server must be registered under the name gsc.

This setup uses suganthan-gsc-mcp, which ships SEO-specific tools beyond raw search analytics (quick_wins, cannibalization_check, inspect_url, content_gaps, …).

// .mcp.json (project) or claude_desktop_config.json (global), under "mcpServers":
"gsc": {
  "command": "npx",
  "args": ["-y", "suganthan-gsc-mcp"],
  "env": {
    "GSC_AUTH_MODE": "oauth",
    "GSC_OAUTH_SECRETS_FILE": "/path/to/your-oauth-client-secret.json",
    "GSC_SITE_URL": "sc-domain:yourdomain.com",
    "GSC_SITE_URLS": "sc-domain:yourdomain.com,sc-domain:yourotherdomain.com"
  }
}

On Windows, wrap the command as "command": "cmd", "args": ["/c", "npx", "-y", "suganthan-gsc-mcp"].

Getting the OAuth secrets file (one-time, ~5 minutes):

Google Cloud Console → create/select a project
Enable the Search Console API
Credentials → Create OAuth client ID (Desktop app) → download the JSON
Point GSC_OAUTH_SECRETS_FILE at it; first run opens a browser to authorize

GSC_SITE_URL is the default property the commands query; GSC_SITE_URLS lets you switch between several properties. Domain properties use the sc-domain: prefix.

Heads-up on date ranges: the tools take a days parameter (28/90/180). Explicit start/end dates are ignored — widen via days.

3. DataForSEO MCP (paid, pay-per-call, required for difficulty/SERP/volume)

The commands call tools named mcp__dataforseo__*, so the server must be registered under the name dataforseo.

"dataforseo": {
  "command": "npx",
  "args": ["dataforseo-mcp-server"],
  "env": {
    "DATAFORSEO_USERNAME": "your-account-email",
    "DATAFORSEO_PASSWORD": "your-api-password"
  }
}

DataForSEO Labs (keyword difficulty, related keywords, keyword ideas, search intent, domain rank)
SERP API (live Google results for validation)

Real-world cost per command run: $0 – $0.18 (each command lists its cost at the bottom; GSC calls are always free). You do NOT need the Backlinks API subscription (~$100+/mo) — nothing here uses it.

4. The knowledge base

The commands cite the methodology by section number ("Section 17", "Section 26.3") — all of it lives in knowledge-base/seo-knowledge-base.md.

Two ways to use it:

Just run the commands. The relevant rules are baked into each command's instructions; the knowledge base is the source they were distilled from.
Give Claude the full base. Copy seo-knowledge-base.md into your project (e.g. docs/) and add one line to your CLAUDE.md: SEO methodology: docs/seo-knowledge-base.md. Now Claude can look up the reasoning behind any rule mid-task — and you can ask it things like "what does the base say about republishing?"

5. Sanity check

Start Claude Code and run:

/seo-quick-wins yourdomain.com

If you get a table of Pos-4-15 keywords (or an honest "0 quick wins"), everything is wired up.

FilesExpand file tree

setup.md

Latest commit

History