Add push monitors for system metrics and fix anthropic version

- Add disk, memory, CPU, and updates push monitor suggestions
- Update Claude prompt to always suggest system health monitors
- Fix anthropic library version compatibility (proxies error)

🤖 Generated with [Claude Code](https://claude.com/claude-code)

Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>

backend/requirements.txt

@@ -2,7 +2,7 @@ flask==3.0.0
 flask-cors==4.0.0
 flask-socketio==5.3.6
 paramiko==3.4.0
-anthropic==0.39.0
+anthropic>=0.40.0
 requests==2.31.0
 python-dotenv==1.0.0
 gevent==23.9.1

backend/services/claude_agent.py

@@ -43,12 +43,13 @@ Always respond with valid JSON in this structure:
     "analysis": "Your analysis of what you found on the host",
     "monitors": [
         {
-            "type": "http|tcp|ping|docker|keyword",
+            "type": "http|tcp|ping|docker|keyword|push",
             "name": "Human-readable monitor name",
-            "target": "URL, hostname, or container name",
+            "target": "URL, hostname, container name, or metric type for push",
             "port": 80,
             "interval": 60,
-            "reason": "Why this should be monitored"
+            "reason": "Why this should be monitored",
+            "push_metric": "disk|memory|cpu|updates (only for push type)"
         }
     ],
     "additional_commands": [
@@ -66,6 +67,31 @@ Always respond with valid JSON in this structure:
 - **ping**: Host availability (provide hostname)
 - **docker**: Docker container status (provide container name)
 - **keyword**: Check for specific text in response (provide URL and keyword)
+- **push**: Push-based monitors for system metrics (scripts on host push to Uptime Kuma)
+
+## Push Monitors for System Metrics
+Always suggest push monitors for system health metrics. These run as cron jobs on the host and push status to Uptime Kuma. Suggest these based on what you see:
+
+1. **Disk Space** - Alert when any partition exceeds 85% usage
+   - Name: "{hostname} - Disk Space"
+   - push_metric: "disk"
+
+2. **Memory Usage** - Alert when RAM exceeds 90% usage
+   - Name: "{hostname} - Memory"
+   - push_metric: "memory"
+
+3. **CPU Load** - Alert when load average exceeds CPU count
+   - Name: "{hostname} - CPU Load"
+   - push_metric: "cpu"
+
+4. **System Updates** - Alert when security updates are pending (Debian/Ubuntu)
+   - Name: "{hostname} - Updates"
+   - push_metric: "updates"
+
+For push monitors, set:
+- type: "push"
+- target: the metric type (disk, memory, cpu, updates)
+- interval: 300 (5 minutes is typical for system metrics)
 
 Be thorough but not excessive. Quality over quantity - suggest monitors that will actually catch real problems."""
 
@@ -81,6 +107,7 @@ class MonitorSuggestion:
     interval: int = 60
     reason: str = ""
     keyword: Optional[str] = None
+    push_metric: Optional[str] = None
 
 
 @dataclass
@@ -234,6 +261,7 @@ Respond with JSON as specified in your instructions."""
                         interval=m.get("interval", 60),
                         reason=m.get("reason", ""),
                         keyword=m.get("keyword"),
+                        push_metric=m.get("push_metric"),
                     )
                 )