Syenite's alert tools let an agent watch lending positions across Aave v3, Morpho Blue, and Spark. When a health factor drops toward a threshold or an LTV approaches a liquidation boundary, the agent can surface a warning and suggest remediation before the position is at risk. This page covers what alerts monitor, how to set them up, and how to forward alert data to a webhook.
Syenite monitors three conditions for registered positions:
| Alert type | Trigger condition | Recommended threshold |
|---|---|---|
| Health factor warning | Health factor falls below threshold | 1.5 (warning) / 1.2 (critical) |
| LTV proximity | LTV within 10% of max liquidation LTV | Protocol-dependent; check max LTV in risk.assess |
| Rate spike | Borrow APR increases by more than N% in 24 hours | 2% absolute change (adjust to position size) |
| Liquidation proximity | Position within 5% of liquidation price | Immediate notification required |
Call alerts.watch with the wallet address, protocol, chain, and health factor threshold. Returns a watch ID.
"tool": "alerts.watch",
"arguments": {
"address": "0xYourWalletAddress",
"protocol": "aave-v3",
"chain": "ethereum",
"healthFactorThreshold": 1.5
}Save the returned watchId — you will use it to poll for alerts and remove the watch when done.
Call alerts.check on a schedule (every 1–5 minutes for active positions). Pass the watch ID to filter results. Set acknowledge: true to clear the alert after processing.
"tool": "alerts.check",
"arguments": {
"watchId": "watch_abc123",
"acknowledge": false
}The response includes alertCount, critical, warning, and an array of alert objects. Each alert has type, severity, message, and data (current health factor, threshold, position details).
The alerts.check response is JSON. Your backend script or agent reads it and forwards to your notification endpoint. Example using curl in a polling script:
#!/bin/bash
RESPONSE=$(curl -s -X POST https://syenite.ai/mcp \
-H "Content-Type: application/json" \
-d '{"jsonrpc":"2.0","method":"tools/call","params":{"name":"alerts.check","arguments":{"watchId":"watch_abc123"}}}')
ALERT_COUNT=$(echo $RESPONSE | jq '.result.content[0].text | fromjson | .alertCount')
if [ "$ALERT_COUNT" -gt "0" ]; then
curl -s -X POST https://your-webhook.example.com/defi-alerts \
-H "Content-Type: application/json" \
-d "$RESPONSE"
fiCall alerts.remove with the watch ID to stop monitoring and free server-side state.
"tool": "alerts.remove",
"arguments": { "watchId": "watch_abc123" }Syenite checks watched positions on each call to alerts.check. The tool does not push alerts independently — your agent or polling script calls alerts.check on a schedule. For active positions, polling every 1–5 minutes is recommended.
Syenite returns alert data as structured JSON from alerts.check. Your agent or backend is responsible for forwarding that data to a webhook endpoint. The JSON payload includes alert type, severity, current health factor, threshold, and position details.
Syenite does not send SMS or email directly. Your integration layer (a backend script, a Zapier workflow, or an agent with notification tools) reads alerts.check output and routes notifications to SMS, email, or Slack as needed.
Syenite alerts are designed to warn before liquidation, not after. If a health factor drops below 1.0, the position is already liquidatable on-chain. Set your healthFactorThreshold to at least 1.3–1.5 to receive warnings with time to act.
Position alerts and carry screening — full alerts workflow including carry screening. Build a DeFi AI agent — how to integrate alert monitoring into a broader agent architecture. Lending rates and risk — how to assess a position before opening it.