Site Analytics — tracking, SEO, and performance audit

You are running the analytics skill. Your job: audit a website's analytics implementation, SEO health, or performance, then help the user pick ONE thing to fix.

Architecture

Layer 1 (deterministic Python): python3 tools/site_analytics.py --root <path> --mode <mode> — scans HTML files for tracking IDs, SEO tags, and performance signals. Generates .tmp/site_analytics_map.json.

Layer 2 (Opus subagent): For seo and full modes, spawn an Explore subagent to check live indexation and competitive signals via WebSearch.

Step-by-step

Step 0 — Parse arguments

Parse $ARGUMENTS for the mode and optional path:

• First word should be one of: tracking, seo, performance, full
• Remaining text is the site root path
• Default mode: full. Default path: current working directory (else ask for the site root)

Step 1 — Run Layer 1

python3 tools/site_analytics.py --root "<site_root>" --mode <mode>

Step 2 — Layer 2 (SEO and full modes only)

For seo or full modes, spawn an Explore subagent:

subagent_type: "Explore"
model: "opus"
description: "Layer 2 SEO analysis"

The subagent should:

1. Read .tmp/site_analytics_map.json
2. Use WebSearch to check: site:<domain> indexation, search for the site's target keywords
3. Assess whether structured data meets Google Rich Results requirements
4. Return findings in the standard format:

[CONCEPTUAL] <category>
    <path>
    → <description>
    Fix: <fix>

Step 3 — Combine and present

Same format as site-audit: numbered findings list, user picks ONE.

Step 4 — Wait, then fix ONE

Same discipline: one fix per invocation.

Mode details

• tracking — GA4/Pixel/Clarity ID consistency, duplicate init, CAPI presence, consent banner
• seo — sitemap, robots.txt, canonical, JSON-LD, title/description quality and uniqueness
• performance — page weight, external requests, render-blocking scripts, image optimization, preconnect
• full — all three combined

See also

• tools/site_analytics.py
• workflows/site_analytics.md