seo-google

Family: seo Status: Stable

Purpose

The "real field data" layer for the SEO family — Google's own measurements of how real users experience a site, versus the synthetic/lab heuristics in seo-technical. It is tool-aware: when a Google Search Console MCP (or Google APIs) is connected it returns true field data; when nothing is connected it still delivers a real on-page technical + GEO read and names precisely what to connect to unlock field data. It never just fails.

What field data adds when connected:

• Search Console — impressions, clicks, CTR, average position; URL Inspection; sitemap status
• PageSpeed Insights / CrUX — real-user Core Web Vitals (LCP, INP, CLS), with CrUX's multi-week history
• GA4 — organic traffic, engagement, conversions

Triggers

• "search console" / "GSC"
• "pagespeed" / "PSI"
• "crux" / "field data" / "real CWV data"
• "indexing api" / "URL inspection"
• "ga4 organic" / "ga4 seo"
• "LCP" / "INP" / "CLS" / "FCP" / "TTFB"
• "impressions" / "clicks" / "CTR" / "position"

Inputs

• Domain or page URL
• Which signal is wanted (search analytics | URL inspection | field CWV | organic traffic)
• Optional: date range / comparison window

Steps

1. Detect available tooling. Run python3 "${CLAUDE_PLUGIN_ROOT}/scripts/workflow/capability_probe.py" (env keys) and check whether a Search Console / Google MCP is exposed this session.
2. Field-data path (Tier 1, if connected). Call the connected GSC/Google MCP or Google APIs for the requested signal; normalize into a compact table; for CrUX, overlay the available history.
3. Built-in path (Tier 2, always available). Fetch the page with WebFetch and run python3 "${CLAUDE_PLUGIN_ROOT}/scripts/seo/tech_audit.py" (on-page technical: title/meta, canonical, robots, viewport, structured-data presence, security headers, INP<200ms guidance) and python3 "${CLAUDE_PLUGIN_ROOT}/scripts/seo/geo_check.py" (AI-citability + llms.txt). This is real, useful output on its own — it just isn't field data.
4. Report which tier ran and, if field data wasn't available, the one-line setup to unlock it (connect a Search Console MCP, or set Google API OAuth).

Capability routing

This skill follows the plugin's capability-tier cascade (references/CAPABILITY-TIERS.md):

1. Tier 1 — Google data. If a Search Console / Google MCP or Google APIs are connected, use them for true field data (GSC analytics, CrUX CWV, GA4 organic).
2. Tier 2 — built-in (default). Otherwise WebFetch + tech_audit.py + geo_check.py for an on-page technical + GEO read. Lab/heuristic, not field.
3. Tier 4 — guided. If a request strictly needs field data (e.g. impressions) and no Google tool is connected, deliver the built-in read and name the exact setup to unlock the field metric.

Always state which tier ran and what connecting Google would add.

Outputs

• The requested signal as a compact table (field data when Tier 1; on-page + GEO findings when Tier 2)
• For CWV: the metric with its target (LCP<2.5s / CLS<0.1 / INP<200ms) and, on Tier 1, real-user values
• A one-line "to unlock field data, connect …" note when running built-in

Dependencies

• Optional (Tier 1): a Google Search Console MCP, or Google APIs via OAuth — adds real field data; free path below works without them
• Built-in (Tier 2): WebFetch; scripts/seo/tech_audit.py; scripts/seo/geo_check.py
• scripts/workflow/capability_probe.py (capability detection)

Notes

When connected, this is the field-data source of truth the rest of the SEO family defers to for "is this actually a problem for real users?" seo-technical covers the synthetic/lab side. Related: seo-geo (AI-citability depth).