SEO Drift Monitor

Protect your SEO work from being broken by page changes. Baseline a page, detect regressions, get fix recommendations.

Commands

Baseline Workflow

When the user runs /seo-drift baseline <url>:

1. Run the baseline capture script:

   python3 ~/.claude/skills/seo-drift/scripts/baseline.py "<url>"
   

2. Parse the JSON output and present a formatted summary:

   Baseline saved for [domain] (ID: [id])
   
   Title: [title]
   H1: [h1 text] ([h1_count] total)
   Schema: [schema_count] blocks ([schema_types])
   Canonical: [canonical]
   Robots: [robots or "none"]
   OG Tags: [og_tags count]
   CWV: [score]/100 (LCP: [lcp]s, FCP: [fcp]s, CLS: [cls])
   

3. If the script returns an error, display it and suggest fixes.

Optional flags:

• Add --skip-cwv to skip the PageSpeed Insights API call (faster)

Check Workflow

When the user runs /seo-drift check <url>:

1. Run the check script:

   python3 ~/.claude/skills/seo-drift/scripts/check.py "<url>"
   

2. Parse the JSON output. If there's an error about no baseline, tell the user to run /seo-drift baseline <url> first.

3. Present diffs grouped by severity:

   Drift Check for [domain]
   Baseline: [baseline_date] | Checked: [checked_at]
   
   CRITICAL ([count]):
   - [element]: [before] -> [after]
     Fix: [recommendation]
   
   WARNING ([count]):
   - [element]: [before] -> [after]
     Fix: [recommendation]
   
   INFO ([count]):
   - [element]: [before] -> [after]
   
   Report: [report_path]
   

4. For CRITICAL and WARNING diffs, cross-reference the appropriate claude-seo skill:

   • Schema issues -> recommend /seo-schema to regenerate
   • CWV regressions -> recommend /seo-technical for diagnosis
   • Content/heading issues -> recommend /seo-page for analysis
   • Meta tag issues -> recommend /seo-page for review

5. If a report was generated, tell the user they can open it in a browser.

History Workflow

When the user runs /seo-drift history <url>:

1. Run:

   python3 ~/.claude/skills/seo-drift/scripts/history.py "<url>"
   

   Replace <url> with the actual URL.

2. Present as a timeline:

   History for [domain]:
   
   Baselines:
   #[id]  [date]  Title: "[title]"  Schema: [count] blocks  CWV: [score]/100
   #[id]  [date]  Title: "[title]"  Schema: [count] blocks  CWV: [score]/100
   
   Checks:
   #[id]  [date] vs baseline #[baseline_id]  [critical]C [warning]W [info]I
   

Severity Classification

Data Storage

• Database: ~/.claude/seo-drift/baselines.db (SQLite)
• Reports: ~/.claude/seo-drift/reports/

Dependencies

This skill requires:

• Python 3.10+
• requests (for HTTP fetching)
• beautifulsoup4 (for HTML parsing)
• curl (for PageSpeed Insights API)

Install: pip install requests beautifulsoup4

Integration with claude-seo

This skill extends the claude-seo ecosystem by adding a defensive monitoring layer. While other seo-* skills optimize pages, seo-drift protects what's already optimized:

• Uses seo/scripts/fetch_page.py for HTTP fetching with SSRF prevention
• Uses seo/scripts/parse_html.py for SEO element extraction
• Recommends seo-schema, seo-technical, and seo-page skills for fixing detected issues