Observability & Metrics

Integrate Acadify telemetry with your existing Prometheus and Grafana stacks.

Prometheus Export

Scrape real-time `/metrics` endpoints for token usage and latency.

Cost Telemetry

Track spend granularly across different evaluation domains.

1. Prometheus Integration

Add the Acadify metrics endpoint directly to your `prometheus.yml` scrape configuration to pull raw operational data.

scrape_configs:
  - job_name: 'acadify_evals'
    scrape_interval: 15s
    metrics_path: '/v1/telemetry/metrics'
    authorization:
      credentials: 'aca_live_abc123'
    static_configs:
      - targets: ['api.acadifysolution.com']

2. Grafana Dashboards

Once Prometheus is scraping our metrics, you can import our official Grafana Dashboard JSON template (ID: 11092) to instantly visualize:

P95 Latency of Execution Sandboxes booting up.
Token Velocity (Tokens per second generated during SWE-bench).
SME Queue Depth (How many RLHF datasets are waiting for human review).