seo-monitor
Skill4 starsChanged 8 months agoMITView on GitHub
SEO monitoring and delta tracking. Re-runs the collector on a previously audited site, compares current data against the baseline from the first audit, and produces a delta report showing improvements, regressions, and alerts. Use this skill for: "monitor my SEO", "check if the fixes worked", "compare my site to the last audit", "run a follow-up audit", "track SEO progress", "has my SEO improved", "re-scan my site", "SEO health check", "check for regressions", "schedule SEO monitoring", "surveille mon SEO", "vérifie les améliorations". Also triggers when the user references a previous audit or asks about SEO progress over time, or when baseline.json exists in the project monitoring/ folder.
2 other projects use a version of this.
Adopted by
Most recently taken up by
- cognyai/claude-code-marketing-skillstheir own version, first seen Oct 12, 2026
- 01clauding/claude-seo-skilltheir own version, first seen Mar 13, 2026
Other versions2 projects use a version of this.
Word for word · 0
None found.
Edited · 0
None found.
Wrote their own version · 2
- cognyai/claude-code-marketing-skills★ 106same name
- 01clauding/claude-seo-skill★ 4same name
How matches are found. Word for word: the exact same text. Edited: shares passages with this file (“99% same” means 99% of its five-word phrases also appear here; 10% or more counts). Their own version: similar meaning, or, weakest, the same name. We can’t tell which project had it first. Projects are counted among the 200 most-starred candidates, and mentions are links to the repository we found, up to 50 per site, so both counts are a floor.
In this repository4 other files it gives its agents.
- seo-collectorSkill
- seo-geo-auditSkill
- seo-implementerSkill
- seo-shared-contextSkill
The file
Show its rules and text
We split this skill into the rules it gives an agent
16 rules. About 0 are common, 11 are distinctive, and 5 are very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈11
Distinctive · estimate
This author's own choices. Worth a look.
5
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Distinctive: line 16
All responses and reports are produced in English by default.
- Very specific: line 64
If no monitoring/baseline.json exists in the project folder:
- Very specific: line 66
Check if site-data.json and audit-actions.json exist from a previous audit
- Distinctive: line 67
If yes → convert them into the baseline:
- Distinctive: line 70
If no → ask the user:
- Very specific: line 81
Run a fresh data collection (re-use seo-collector methodology)
- Distinctive: line 106
If Ahrefs MCP was used in the baseline collection, re-query the same endpoints for delta comparison:
- Distinctive: line 325
After generating the monitoring report, present:
- Distinctive: line 364
When running as a scheduled task (Pattern 3 from the skill system architecture):
- Distinctive: line 387
Run the quick scan
- Very specific: line 390
If all clear, log the scan silently to monitoring/history.md
- Very specific: line 413
Never overwrite baseline.json — it's the reference point
- Distinctive: line 415
If a full re-audit is done, offer to update the baseline (with user confirmation)
- Distinctive: line 418
Don't alert on changes within normal variance (< 2% indexation change, < 0.5pt score change)
- Distinctive: line 423
Always compare against the same data points
- Distinctive: line 424
If the collector methodology changed between scans, note it in the delta report
≈ “Common” is an estimate: at least 1 in 5 of the 372,429 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.
Files
Skill: SEO Monitor — Periodic Delta Tracking
All responses and reports are produced in English by default. Switch to the user's language if they explicitly request it.
Overview
This skill is Step 4 in the SEO-GEO pipeline. It re-crawls a previously audited site, compares the current state against the original baseline, and produces a delta report highlighting what improved, what regressed, and what needs attention.
Position in the pipeline
[seo-collector] → site-data.json ──────────────────────┐
↓ │
[seo-geo-audit] → audit-report.md + audit-actions.json │
↓ │
[seo-implementer] → fixes/ │
↓ │
┌─────────────────────────┐ │
│ SEO MONITOR (this) │ ←──────────────┘
└─────────────────────────┘ (re-uses collector)
↓
monitoring/
├── baseline.json
├── delta-[date].md
└── alerts.md
↓
↻ loops back to collector for next cycleWhat it produces
seo-project-[domain]/
└── monitoring/
├── baseline.json ← Snapshot from the first audit (created once)
├── scan-[date].json ← Each new crawl result
├── delta-[date].md ← Comparison report (human-readable)
├── delta-[date].json ← Comparison data (machine-readable)
├── alerts.md ← Active alerts (updated each scan)
└── history.md ← Score evolution over timeStep 0: Determine monitoring mode
First run (no baseline exists)
If no monitoring/baseline.json exists in the project folder:
- Check if
site-data.jsonandaudit-actions.jsonexist from a previous audit - If yes → convert them into the baseline:
- Copy
site-data.json→monitoring/baseline.json(addbaseline_datefield) - Extract scores from
audit-actions.json→ store in baseline
- Copy
- If no → ask the user:
No previous audit found for this domain. Would you like me to: 1. Run a full audit first (collector + audit) to establish the baseline 2. Run a quick scan now and use it as the baseline
Subsequent runs (baseline exists)
- Load
monitoring/baseline.json - Run a fresh data collection (re-use
seo-collectormethodology) - Compare and generate the delta
Step 1: Re-collect current data
Re-run the collector methodology on the same URL. Collect the same data points as the original site-data.json:
1.1 Quick scan (default)
Focused on the metrics most likely to change:
| Data point | Method | Why it matters |
| Homepage meta tags | web_fetch homepage | Detect if fixes were applied |
| Robots.txt | web_fetch /robots.txt | Detect if AI bots were unblocked |
| Core Web Vitals | web_search PageSpeed data | Detect performance changes |
| Indexation count | web_search site:[domain] | Detect indexation gains/losses |
| Schema markup | web_fetch HTML check | Detect if schemas were added |
| Sample pages (3-5) | web_fetch key pages | Detect meta tag changes |
| AI bot access | Parse robots.txt for each bot | Detect GEO improvements |
Ahrefs MCP data points (if connected)
If Ahrefs MCP was used in the baseline collection, re-query the same endpoints for delta comparison:
| Ahrefs data point | Endpoint | What to compare |
| Domain Rating | domain-rating | DR change since baseline (gaining/losing authority?) |
| Referring domains | backlinks-stats | New referring domains acquired? Lost any? |
| Broken backlinks | broken-backlinks | Were broken links fixed? New ones appeared? |
| Organic keywords | organic-keywords | Keyword position changes, new rankings gained |
| Top pages traffic | top-pages | Traffic shifts on key pages |
| Brand Radar | brand-radar | AI citation changes (new mentions? lost mentions?) |
Brand Radar delta is critical for GEO monitoring: if the baseline showed 3 AI citations and the current scan shows 5, that's a direct measure of GEO improvement. If it dropped, investigate why (content freshness? bot access changed?).
1.2 Full re-scan (on request)
Runs the complete seo-collector methodology. Use when the user requests a comprehensive re-audit or when significant changes are expected.
Save the scan result to monitoring/scan-[date].json.
Step 2: Compare against baseline
For each data point, compute the delta:
2.1 Score comparison
{
"comparison_date": "[ISO 8601]",
"baseline_date": "[ISO 8601 from baseline]",
"days_since_baseline": 0,
"scores": {
"technical_seo": {
"baseline": 0,
"current": 0,
"delta": 0,
"trend": "improved|regressed|unchanged"
},
"core_web_vitals": { "..." },
"local_or_content": { "..." },
"eeat": { "..." },
"entity_seo": { "..." },
"geo": { "..." },
"ux_conversion": { "..." },
"overall": {
"baseline": 0,
"current": 0,
"delta": 0,
"max": 90
}
}
}2.2 Specific change detection
For each category, track specific changes:
Meta tags changes:
{
"url": "https://example.com/page",
"field": "title",
"baseline_value": "[old title]",
"current_value": "[new title]",
"status": "fixed|regressed|unchanged|new_issue"
}Robots.txt changes:
{
"bot": "GPTBot",
"baseline_status": "blocked",
"current_status": "allowed",
"status": "fixed"
}Indexation changes:
{
"google_indexed": { "baseline": "~340", "current": "~380", "delta": "+~40" },
"bing_indexed": { "baseline": true, "current": true },
"brave_indexed": { "baseline": "unknown", "current": true }
}Core Web Vitals changes:
{
"lcp_mobile": { "baseline": "3.2s", "current": "2.4s", "delta": "-0.8s", "status": "improved" },
"cls_mobile": { "baseline": "0.12", "current": "0.08", "delta": "-0.04", "status": "improved" },
"inp_mobile": { "baseline": "280ms", "current": "190ms", "delta": "-90ms", "status": "improved" }
}Save the full comparison to monitoring/delta-[date].json.
Step 3: Generate alerts
Evaluate each change against alert thresholds:
Alert levels
| Level | Trigger | Example |
| 🔴 CRITICAL | Score dropped by > 10 points overall, or a previously fixed issue has returned | Overall score went from 62 to 48; H1 that was fixed is missing again |
| 🟠 WARNING | Score dropped in any single dimension by > 3 points, or a new issue appeared | GEO score dropped from 14 to 10; a new 404 page detected |
| 🟢 POSITIVE | Score improved in any dimension, or a fix was confirmed working | Core Web Vitals improved; schema markup now detected |
| ℹ️ INFO | Change detected but no clear positive/negative impact | Indexation count changed by < 5%; new page added to sitemap |
Alert file: monitoring/alerts.md
# SEO Alerts — [domain]
Last scan: [date] | Baseline: [date] | Days elapsed: [N]
## 🔴 CRITICAL
- [Description of critical alert with details and recommended action]
## 🟠 WARNING
- [Description of warning with details]
## 🟢 POSITIVE CHANGES
- [Description of improvement confirmed]
## ℹ️ INFO
- [Description of neutral change]
---
No alerts = ✅ All clear — site is stable since last scan.Step 4: Generate delta report
File: monitoring/delta-[date].md
# SEO Monitoring Report — [domain]
Scan date: [date] | Baseline: [date] | Days since baseline: [N]
---
## Score evolution
| Dimension | Baseline | Current | Delta | Trend |
|---|---|---|---|---|
| Technical SEO | [X]/10 | [Y]/10 | [+/-Z] | ↑/↓/→ |
| Core Web Vitals | [X]/10 | [Y]/10 | [+/-Z] | ↑/↓/→ |
| Local/Content & Authority | [X]/20 | [Y]/20 | [+/-Z] | ↑/↓/→ |
| E-E-A-T | [X]/10 | [Y]/10 | [+/-Z] | ↑/↓/→ |
| Entity SEO | [X]/10 | [Y]/10 | [+/-Z] | ↑/↓/→ |
| GEO | [X]/20 | [Y]/20 | [+/-Z] | ↑/↓/→ |
| UX & Conversion | [X]/10 | [Y]/10 | [+/-Z] | ↑/↓/→ |
| **OVERALL** | **[X]/90** | **[Y]/90** | **[+/-Z]** | **↑/↓/→** |
## Fixes confirmed working ✅
[List each fix from audit-actions.json that has been successfully implemented]
## Fixes not yet applied ⏳
[List each fix that was recommended but hasn't been implemented yet]
## Regressions detected 🔴
[List any metrics that got worse since the baseline]
## New issues found 🆕
[List any problems that didn't exist at baseline time]
## Recommendations
[Prioritized next actions based on the current state]
---
## Next scan recommended: [date — baseline + interval]Step 5: Update history
File: monitoring/history.md
Append each scan result to a running history:
# SEO Score History — [domain]
| Date | Overall /90 | Tech | CWV | Content | E-E-A-T | Entity | GEO | UX | Notes |
|---|---|---|---|---|---|---|---|---|---|
| [baseline date] | [X] | [x] | [x] | [x] | [x] | [x] | [x] | [x] | Initial audit |
| [scan date] | [Y] | [y] | [y] | [y] | [y] | [y] | [y] | [y] | [key change noted] |
| ... | | | | | | | | | |This table grows with each scan. It provides a quick view of the site's SEO health trajectory.
Monitoring frequency recommendations
| Site type | Recommended interval | Rationale |
| E-commerce (active) | Weekly | Frequent product/price changes, competitive market |
| SaaS / B2B | Bi-weekly | Moderate content velocity, feature pages change |
| Blog / Media | Weekly | New content published frequently |
| Local business | Monthly | Slower pace of change, review monitoring more important |
| Corporate / Institutional | Monthly | Rarely changes, but regressions must be caught |
| Post-fix verification | 1 week after deployment | Confirm fixes took effect before next audit |
Output summary
After generating the monitoring report, present:
═══════════════════════════════════════════
SEO MONITORING REPORT — [domain]
Scan: [date] | vs Baseline: [date] ([N] days)
═══════════════════════════════════════════
📊 Score evolution: [baseline]/90 → [current]/90 ([+/-delta])
✅ Fixes confirmed: [N] of [total] recommendations implemented
⏳ Pending fixes: [N] not yet applied
🔴 Regressions: [N] metrics worsened
🆕 New issues: [N] new problems detected
📁 Files updated:
monitoring/delta-[date].md — Full comparison report
monitoring/delta-[date].json — Machine-readable delta
monitoring/alerts.md — Active alerts
monitoring/history.md — Score timeline
[If regressions detected:]
⚠️ Action needed: [brief description of most critical regression]
[If all good:]
✅ All stable — no critical changes since baseline.
Next scan recommended: [date]
Would you like me to:
1. Detail the regressions and suggest fixes?
2. Re-run the full audit for a complete re-score?
3. Generate an updated implementation checklist?Integration with scheduled tasks
When running as a scheduled task (Pattern 3 from the skill system architecture):
Cron / scheduled task setup
# Weekly SEO monitoring — runs every Monday at 7am
0 7 * * 1 claude "Run SEO monitor on https://[domain]"
# Bi-weekly
0 7 * * 1 [ $(( $(date +%V) % 2 )) -eq 0 ] && claude "Run SEO monitor on https://[domain]"
# Monthly (1st of month)
0 7 1 * * claude "Run SEO monitor on https://[domain]"Claude Code / Cowork scheduled tasks
Schedule: "Every Monday at 7am, run a quick SEO scan on https://[domain] and compare to baseline. If any critical alerts, notify me."The monitor should:
- Load the baseline silently
- Run the quick scan
- Generate the delta
- Only notify the user if there are 🔴 CRITICAL or 🟠 WARNING alerts
- If all clear, log the scan silently to
monitoring/history.md
Pattern 3 applied: Scheduled orchestration. The monitor runs on a schedule without user intervention. It triggers alerts only when something needs attention. The user reviews results, not processes.
Edge cases
| Situation | Handling |
| Site is down during scan | Log the failure, retry once after 5 min. If still down → 🔴 CRITICAL alert "Site unreachable" |
| Domain changed | Detect via redirect. Ask user to confirm new domain. Create new baseline. |
| Major redesign detected | If > 50% of tracked URLs return 404 or different structure → 🟠 WARNING "Major site changes detected — recommend full re-audit" |
| CMS changed | Detect via cms_detected mismatch. Flag in delta report. |
| robots.txt disappeared | 🔴 CRITICAL alert — all crawlers may be blocked |
| Score dropped but fixes were applied | Check if fixes are still in place (could be overwritten by CMS update). Flag specifically. |
| Baseline is > 6 months old | ℹ️ INFO "Baseline is aging — consider running a full re-audit to reset" |
Best practices
Data continuity
- Never overwrite
baseline.json— it's the reference point - Each scan creates a new
scan-[date].json— full history preserved - If a full re-audit is done, offer to update the baseline (with user confirmation)
Alert fatigue prevention
- Don't alert on changes within normal variance (< 2% indexation change, < 0.5pt score change)
- Group related alerts (if 3 meta tags regressed on the same template, that's 1 alert, not 3)
- Distinguish between "fix not applied yet" (expected) and "fix was applied but broke" (critical)
Comparison accuracy
- Always compare against the same data points
- If the collector methodology changed between scans, note it in the delta report
- Estimations vs. real data: if baseline was estimated and current has real data, flag the methodology difference
Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
No reviews yet
People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.
- It worked
- Partly
- It didn't
Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.
Your trail
Files you open will collect here, so you can find them again.
Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.
Now do this for your own.
Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

