feat: Integrate GA4 analytics into SEO agent

- Added GA4 traffic monitoring to seo-agent.py
- Tracks sessions, users, bounce rate from GA4
- Detects traffic anomalies (>50% drop triggers alert)
- Maintains 30-day traffic history in state
- Updated daily-report.sh with enhanced GA4 metrics
- GA4 data now flows to morning brief
- Hourly checks every 6 hours to avoid API fatigue

agents/marketing-seo/scripts/daily-report.sh

@@ -1,42 +1,67 @@
 #!/bin/bash
 # Daily SEO Report - 8 AM UTC
+# Includes GA4 Analytics Data
 
 WORKSPACE="/Users/claw/.openclaw/workspace/agents/marketing-seo"
 LOG="$WORKSPACE/logs"
-
 cd $WORKSPACE
 
 # Get GA4 data
-GA=$(python3 scripts/ga4-direct.py 2>/dev/null | grep -A3 "Traffic Data")
+echo "📊 Fetching GA4 Analytics..."
-SESSIONS=$(echo "$GA" | grep Sessions | grep -o "[0-9]*")
+GA_OUTPUT=$(python3 scripts/ga4-direct.py 2>/dev/null)
-USERS=$(echo "$GA" | grep Users | grep -o "[0-9]*")
+
+# Parse GA4 metrics
+SESSIONS=$(echo "$GA_OUTPUT" | grep -i "Sessions:" | grep -o "[0-9]*" | head -1)
+USERS=$(echo "$GA_OUTPUT" | grep -i "Users:" | grep -o "[0-9]*" | head -1)
+BOUNCE=$(echo "$GA_OUTPUT" | grep -i "Bounce Rate:" | grep -o "[0-9.]*" | head -1)
+
+# Set defaults if empty
+SESSIONS=${SESSIONS:-"N/A"}
+USERS=${USERS:-"N/A"}
+BOUNCE=${BOUNCE:-"N/A"}
 
 # Get site status
 WWW_UP=$(curl -s -o /dev/null -w "%{http_code}" https://www.hoaledgeriq.com -m 10)
 APP_UP=$(curl -s -o /dev/null -w "%{http_code}" https://app.hoaledgeriq.com -m 10)
 
-# Rankings
+# Format status icons
-RANK_STATUS="Establishment phase (not yet in top 100)"
+WWW_ICON="✅"
+APP_ICON="✅"
+if [ "$WWW_UP" != "200" ]; then WWW_ICON="❌"; fi
+if [ "$APP_UP" != "200" ]; then APP_ICON="❌"; fi
 
-# Send Telegram report
+# Get rankings status (from rank-tracker if available)
+RANK_FILE="$WORKSPACE/state/rank-data.json"
+if [ -f "$RANK_FILE" ]; then
+    KEYWORDS=$(cat "$RANK_FILE" | grep -o '"keywords":\[[^]]*\]' | grep -o '[0-9]*' | head -1)
+    RANK_STATUS="Tracking $KEYWORDS keywords"
+else
+    RANK_STATUS="Baseline monitoring active"
+fi
+
+# Build message
 MSG="📊 *DAILY SEO REPORT* - $(date '+%a %b %d')
 
 🌐 *Sites:*
-✅ www.hoaledgeriq.com: ${WWW_UP}
+${WWW_ICON} www.hoaledgeriq.com: ${WWW_UP}
-✅ app.hoaledgeriq.com: ${APP_UP}
+${APP_ICON} app.hoaledgeriq.com: ${APP_UP}
 
 📈 *Traffic (24h):*
-• Sessions: ${SESSIONS:-0}
+• Sessions: ${SESSIONS}
-• Users: ${USERS:-0}
+• Users: ${USERS}
+• Bounce Rate: ${BOUNCE}%
 
 📈 *Rankings:*
-${RANK_STATUS}
+• ${RANK_STATUS}
-• 8 keywords tracked
+• Monitoring for breakthrough
-• Baseline established
-• Monitoring for break-through
 
-⚡ Status: Healthy ✅"
+⚡ Status: Healthy ✅
 
+_GA4 Analytics Integrated_"
+
+# Send via Telegram
 openclaw message send --channel telegram --target telegram:8269921691 --message "$MSG" 2>/dev/null || echo "$MSG" >> "$LOG/daily-$(date +%Y%m%d).log"
 
+# Log success
 echo "Report sent: $(date)" >> "$LOG/report-sent.log"
+echo "Daily report completed at $(date)"

agents/marketing-seo/scripts/seo-agent.py

@@ -3,26 +3,30 @@
 Marketing-SEO Agent - 24/7 Continuous Monitoring
 Monitors: site health, rankings, traffic, competitors
 Alerts: Telegram/email on critical issues
+GA4 Integration: Tracks traffic anomalies, session drops, user engagement
 """
 import json
 import time
 import urllib.request
-from datetime import datetime
+from datetime import datetime, timedelta
 from pathlib import Path
 import subprocess
+import sys
 
 WORKSPACE = Path(__file__).parent.parent
 LOG_DIR = WORKSPACE / "logs"
 STATE_FILE = WORKSPACE / "state" / "agent-state.json"
 CONFIG_FILE = WORKSPACE / "config" / "agent-config.yaml"
-
+GA4_SCRIPT = WORKSPACE / "scripts" / "ga4-direct.py"
 LOG_DIR.mkdir(parents=True, exist_ok=True)
 
 SITES = [
     "https://www.hoaledgeriq.com",
     "https://app.hoaledgeriq.com"
 ]
+
 MONITOR_INTERVAL = 3600  # 1 hour
+TRAFFIC_DROP_THRESHOLD = 0.50  # Alert if traffic drops >50%
 
 def log(msg):
     ts = datetime.now().strftime('%Y-%m-%d %H:%M:%S')
@@ -34,7 +38,7 @@ def log(msg):
 def load_state():
     if STATE_FILE.exists():
         return json.loads(STATE_FILE.read_text())
-    return {"last_check": None, "alerts_today": 0, "status": "running"}
+    return {"last_check": None, "alerts_today": 0, "status": "running", "traffic_history": []}
 
 def save_state(s):
     STATE_FILE.write_text(json.dumps(s, indent=2))
@@ -49,36 +53,75 @@ def check_site_health(url):
     except Exception as e:
         return False, str(e), None
 
-def run_seo_audit():
+def get_ga4_data():
-    """Run basic SEO checks using web tools"""
+    """
-    results = {
+    Get GA4 traffic data from ga4-direct.py
-        "site_up": False,
+    Returns: dict with sessions, users, bounce_rate or None if error
-        "response_time": None,
+    """
-        "ssl_valid": True,
-        "robots_accessible": False,
-        "sitemap_exists": False
-    }
-    
-    # Check main site
-    start = time.time()
-    results["site_up"], status = check_site_health()
-    results["response_time"] = round(time.time() - start, 2)
-    
-    # Check robots.txt
     try:
-        urllib.request.urlopen(f"{SITE_URL}/robots.txt", timeout=5)
+        result = subprocess.run(
-        results["robots_accessible"] = True
+            [sys.executable, str(GA4_SCRIPT)],
-    except:
+            capture_output=True,
-        pass
+            text=True,
+            timeout=30
+        )
         
-    # Check sitemap
+        if result.returncode == 0:
-    try:
+            # Parse output - looks for lines like "Sessions: 123"
-        urllib.request.urlopen(f"{SITE_URL}/sitemap.xml", timeout=5)
+            output = result.stdout
-        results["sitemap_exists"] = True
+            data = {}
-    except:
-        pass
             
-    return results
+            for line in output.split('\n'):
+                if 'Sessions:' in line:
+                    data['sessions'] = int(''.join(filter(str.isdigit, line.split('Sessions:')[1])))
+                elif 'Users:' in line or 'Active Users:' in line:
+                    data['users'] = int(''.join(filter(str.isdigit, line.split('Users:')[1])))
+                elif 'Bounce Rate:' in line:
+                    data['bounce_rate'] = float(''.join(filter(lambda x: x.isdigit() or x == '.', line.split('Bounce Rate:')[1])))
+            
+            return data if data else None
+    except Exception as e:
+        log(f"GA4 fetch error: {e}")
+        return None
+
+def check_traffic_anomalies(current_data):
+    """
+    Check if current traffic has significant drops compared to historical data
+    Returns: alert message if anomaly detected, None otherwise
+    """
+    state = load_state()
+    history = state.get('traffic_history', [])
+    
+    if not current_data or 'sessions' not in current_data:
+        return None
+    
+    current_sessions = current_data.get('sessions', 0)
+    
+    # Need at least 2 days of history to compare
+    if len(history) >= 2:
+        avg_sessions = sum(h.get('sessions', 0) for h in history[-7:]) / min(len(history), 7)
+        
+        if avg_sessions > 0:
+            drop_pct = (avg_sessions - current_sessions) / avg_sessions
+            
+            if drop_pct > TRAFFIC_DROP_THRESHOLD:
+                return f"🚨 Traffic drop detected: {drop_pct*100:.1f}% below average\n• Current: {current_sessions} sessions\n• 7-day avg: {avg_sessions:.0f} sessions\n• Threshold: {TRAFFIC_DROP_THRESHOLD*100}% drop"
+    
+    # Add to history (keep last 30 days)
+    history.append({
+        'date': datetime.now().strftime('%Y-%m-%d'),
+        'sessions': current_sessions,
+        'users': current_data.get('users', 0),
+        'timestamp': datetime.now().isoformat()
+    })
+    
+    if len(history) > 30:
+        history = history[-30:]
+    
+    state['traffic_history'] = history
+    save_state(state)
+    
+    return None
 
 def send_alert(title, message, severity="warning"):
     """Send alert via multiple channels"""
@@ -87,65 +130,87 @@ def send_alert(title, message, severity="warning"):
     # Telegram alert
     try:
         tg_msg = f"🔔 *SEO Alert: {title}*\n\n{message}\n\n⏰ {datetime.now().strftime('%H:%M')}"
-        subprocess.run(["openclaw", "message", "send", "--text", tg_msg], 
+        subprocess.run(["openclaw", "message", "send", "--text", tg_msg], capture_output=True, timeout=10)
-                      capture_output=True, timeout=10)
+    except Exception as e:
-    except:
+        log(f"Telegram send failed: {e}")
-        pass
     
-    # Log to alerts
+    # Log to alerts file
     with open(LOG_DIR / f"alerts-{datetime.now().strftime('%Y%m%d')}.log", 'a') as f:
         f.write(f"[{severity.upper()}] {datetime.now().isoformat()}: {title}\n{message}\n\n")
 
 def hourly_check():
-    """Run every hour - check both sites"""
+    """Run every hour - check both sites and GA4 traffic"""
     log("=== Hourly Site Check ===")
     
-    all_healthy = True
+    results = {
-    results = {}
+        'sites': {},
+        'traffic': None,
+        'timestamp': datetime.now().isoformat()
+    }
     
+    # Check site health
     for site in SITES:
         log(f"Checking {site}...")
         is_up, status, response_time = check_site_health(site)
-        results[site] = {"up": is_up, "status": status, "time": response_time}
+        results['sites'][site] = {"up": is_up, "status": status, "time": response_time}
         
         if is_up:
             log(f"✅ {site}: UP ({status}) - {response_time}s")
         else:
             log(f"❌ {site}: DOWN ({status})")
-            send_alert(f"SITE DOWN: {site}", f"Status: {status}", "critical")
+            send_alert(f"SITE DOWN: {site}", f"Status: {status}\nURL: {site}", "critical")
-            all_healthy = False
+    
+    # Check GA4 traffic (every 6 hours to avoid API fatigue)
+    if datetime.now().hour % 6 == 0:
+        log("Fetching GA4 traffic data...")
+        traffic_data = get_ga4_data()
+        
+        if traffic_data:
+            results['traffic'] = traffic_data
+            log(f"📊 GA4 Data: {traffic_data.get('sessions', 0)} sessions, {traffic_data.get('users', 0)} users")
+            
+            # Check for anomalies
+            anomaly = check_traffic_anomalies(traffic_data)
+            if anomaly:
+                send_alert("Traffic Anomaly Detected", anomaly, "warning")
+        else:
+            log("⚠️ Could not fetch GA4 data")
     
     return results
 
 def daily_report():
-    """Generate daily summary"""
+    """Generate comprehensive daily summary with GA4 data"""
     log("=== Daily SEO Report ===")
     
-    # Compile stats
+    # Get GA4 data
-    s = load_state()
+    traffic_data = get_ga4_data()
+    traffic_report = ""
     
-    # Check Search Console (if configured)
+    if traffic_data:
-    # This would integrate with actual APIs
+        traffic_report = f"""
+📈 *Traffic (24h):*
+• Sessions: {traffic_data.get('sessions', 'N/A')}
+• Users: {traffic_data.get('users', 'N/A')}
+• Bounce Rate: {traffic_data.get('bounce_rate', 'N/A')}%"""
+    else:
+        traffic_report = "\n📈 *Traffic:* Data unavailable"
     
     report = f"""📊 SEO Daily Report - {datetime.now().strftime('%Y-%m-%d')}
 
-Site Status: ✅ Healthy
+🌐 *Site Status:*
-Response Time: ~200ms
+• www.hoaledgeriq.com: ✅ UP
-SSL: Valid
+• app.hoaledgeriq.com: ✅ UP
-Monitoring: 24/7 Active
+{traffic_report}
 
-Tomorrow's Focus:
+⚡ *Status:* Healthy
-- Competitor analysis
+🔍 *Focus:* Competitor analysis & rankings"""
-- Rankings check  
-- Content opportunities
-
-No critical issues detected."""
 
     send_alert("Daily SEO Summary", report, "info")
 
 def main():
-    log("🚀 Marketing-SEO Agent Started - Hourly Mode")
+    log("🚀 Marketing-SEO Agent Started - Hourly Mode with GA4 Integration")
     log(f"Monitoring: {', '.join(SITES)}")
+    log(f"GA4 Script: {GA4_SCRIPT}")
     
     last_check = 0
     last_daily = None
@@ -160,7 +225,7 @@ def main():
             last_check = now_ts
         
         # Daily report at 08:00
-        if now.hour == 8 and now.strftime('%Y-%m-%d') != last_daily:
+        if now.hour == 8 and now.minute == 0 and now.strftime('%Y-%m-%d') != last_daily:
             daily_report()
             last_daily = now.strftime('%Y-%m-%d')