MotteAW provides comprehensive monitoring, debugging, and analytics for multi-agent systems with real-time streaming interfaces and historical performance analysis.
Real-time streaming of agent activities, API calls, and system events with WebSocket connections.
Comprehensive analytics with customizable dashboards and historical trend analysis.
Intelligent alerting with configurable thresholds, notification channels, and escalation policies.
Advanced debugging capabilities with request/response inspection, trace analysis, and replay functionality.
| Metric Type | Description | Update Frequency | Retention |
|---|---|---|---|
| Response Time | Agent execution latency | Real-time | 30 days |
| Success Rate | Successful vs failed requests | 1 minute | 90 days |
| Token Usage | API token consumption | Real-time | 1 year |
| Error Rate | Error frequency by type | Real-time | 90 days |
| Memory Usage | Memory bank utilization | 5 minutes | 30 days |
{
"alert_name": "High Error Rate",
"condition": {
"metric": "error_rate",
"operator": "greater_than",
"threshold": 0.05, // 5%
"window": "5m"
},
"notifications": [
{
"type": "email",
"recipients": ["admin@company.com"]
},
{
"type": "slack",
"channel": "#alerts"
}
],
"severity": "high"
}Configure alerts based on metric thresholds with multiple notification channels.
{
"alert_name": "Response Time Anomaly",
"condition": {
"type": "anomaly",
"metric": "response_time",
"sensitivity": "medium",
"baseline_window": "7d",
"detection_window": "1h"
},
"auto_resolve": true,
"cooldown": "30m"
}Machine learning-based anomaly detection for unusual patterns in agent behavior.
Complete request lifecycle tracking from input to output with detailed timing information and intermediate steps.
Detailed inspection of API responses, model outputs, and data transformations with syntax highlighting.
Replay historical requests for debugging and testing with the ability to modify parameters and compare results.
Detailed performance analysis with bottleneck identification and optimization recommendations.
GET /api/monitoring/metrics?metric=response_time&window=1h&granularity=1m
Response:
{
"metric": "response_time",
"window": "1h",
"data": [
{"timestamp": "2024-01-15T10:00:00Z", "value": 245.5},
{"timestamp": "2024-01-15T10:01:00Z", "value": 238.2},
...
],
"summary": {
"avg": 242.1,
"min": 198.3,
"max": 312.7,
"p95": 289.4
}
}Retrieve historical metrics with configurable time windows and granularity.
POST /api/monitoring/alerts
Content-Type: application/json
{
"name": "High Response Time",
"condition": {
"metric": "response_time",
"operator": "greater_than",
"threshold": 1000,
"window": "5m"
},
"notifications": [
{"type": "email", "recipients": ["admin@company.com"]}
]
}Create custom alerts with flexible conditions and notification preferences.
// Connect to real-time stream
const ws = new WebSocket('wss://api.motte.ai/monitoring/stream');
ws.onmessage = (event) => {
const data = JSON.parse(event.data);
console.log('Real-time update:', data);
};
// Subscribe to specific metrics
ws.send(JSON.stringify({
action: 'subscribe',
metrics: ['response_time', 'error_rate', 'token_usage']
}));Real-time streaming of metrics and events via WebSocket connections.