See everything your AI does
Complete visibility into every tool call, every cost, and every error. Make data-driven decisions about your AI infrastructure.
Real-time dashboards
See requests, costs, and errors as they happen. No delay, no sampling.
Cost attribution
Know exactly which tools, users, and projects drive your spending.
Anomaly detection
Get alerted when usage patterns deviate from normal behavior.
Insights that matter
Understand your patterns
See which tools are called most, when peak usage occurs, and how patterns change over time.
Query your data programmatically
Export analytics via API for custom dashboards or alerting.
import { Primrose } from 'primrose-mcp'
const primrose = new Primrose({ apiKey: process.env.PRIMROSE_KEY })
// Get usage stats for the last 7 days
const usage = await primrose.analytics.usage({
period: '7d',
groupBy: 'tool',
})
// Get cost breakdown by project
const costs = await primrose.analytics.costs({
period: 'current_month',
groupBy: 'project',
})
// Set up alerts
await primrose.analytics.createAlert({
name: 'High error rate',
condition: 'error_rate > 5%',
window: '15m',
notify: ['ops@company.com'],
})Complete observability
Request timeline
See every request with timing, status, and full context.
Latency percentiles
Track p50, p95, p99 latencies for every tool.
Custom alerts
Set thresholds for usage, cost, errors, or latency.
Export & integrate
Export to CSV or send to Datadog, Grafana, or your SIEM.
Request inspection
Drill into any request to see full input and output.
Retention controls
Choose how long to retain detailed logs for compliance.
Get visibility today
Analytics are included in every plan. Start seeing your data immediately.