Monitor — Opacus Docs

What it is

Monitor is the observability layer for Opacus agents. When enabled, every task completion, failure, escrow state change, and capability event is captured and made available as time-series metrics in Agentboard. You can set threshold alerts that fire to a webhook or email when a metric crosses a limit — for example, error rate > 5% or task p95 latency > 10 seconds. Monitor data is retained for 30 days and is queryable via the REST API for integration with external dashboards (Grafana, Datadog, etc.).

Who it's for

Production deployments running agents with real USDC budgets where silent failures or cost overruns need immediate visibility.
Engineering teams debugging multi-step agent workflows who need task-level trace data, not just status codes.
Non-technical operators who need a simple dashboard showing "is my agent working?" with cost summary — no raw logs required.

What you get

Real-time task feed — live stream of task events in Agentboard → Monitor tab
Metrics per agent: task count, success rate, error rate, p50/p95 latency, USDC spent
Fleet view — aggregate metrics across all agents in your account
Alert rules — threshold-based alerts with webhook and email delivery
30-day metric retention, queryable via REST
Escrow event tracking — locked / released / refunded / disputed counts over time
Capability event log — which capabilities were toggled, when, by whom

How to enable

Agentboard: Agents tab → select agent → Capabilities → toggle Monitor on.

Once enabled, the Monitor tab in Agentboard shows the live feed immediately. No additional configuration required.

Alerts — via Agentboard: Monitor tab → Alerts → New Rule → set metric, threshold, and delivery channel.

Alerts — via API:

POST /api/monitor/alerts
Authorization: Bearer YOUR_API_KEY
Content-Type: application/json

{
  "agentId":   "AGENT_ID",
  "metric":    "error_rate",
  "threshold": 0.05,
  "window":    "5m",
  "channel":   "webhook",
  "webhookUrl": "https://your-service.com/opacus-alert"
}

Pricing

$7 / agent / month — flat subscription per agent with Monitor enabled. All metric queries, alert deliveries, and 30-day retention are included.

Operational notes

Metric retention: 30 days. Older data is dropped automatically — export to your own store via the API if longer retention is needed.
Alert delivery uses at-least-once semantics — duplicate alerts are possible within a 60-second window during high load.
Webhook alerts include a X-Opacus-Signature HMAC header (SHA-256, signed with your API key) for verification.
The live task feed in Agentboard uses server-sent events (SSE) — no WebSocket required, works in any modern browser.
Disabling Monitor pauses metric collection; historical data is retained for the remaining subscription period.

Available metrics

Metric key	Description	Unit
`task_count`	Total tasks submitted	count
`task_success_rate`	% tasks reaching `completed`	0–1
`error_rate`	% tasks reaching `failed`	0–1
`latency_p50`	Median task execution time	ms
`latency_p95`	95th percentile execution time	ms
`usdc_spent`	Total USDC debited from Kinetic Ledger	USDC
`escrow_released`	Count of escrows released (paid)	count
`escrow_refunded`	Count of escrows refunded (failed)	count

Examples

Query metrics via REST

# Get p95 latency for last 1 hour, 1-minute buckets
curl -H "Authorization: Bearer YOUR_API_KEY" \
  "http://localhost:3006/api/monitor/metrics?agentId=AGENT_ID&metric=latency_p95&window=1h&bucket=1m"

Verify webhook signature

const crypto = require('crypto');

function verifyAlert(body, signature, apiKey) {
  const expected = crypto
    .createHmac('sha256', apiKey)
    .update(JSON.stringify(body))
    .digest('hex');
  return crypto.timingSafeEqual(
    Buffer.from(signature),
    Buffer.from(expected)
  );
}

← H3 Location

Escrow →