Analytics Guide

Choose a metrics backend and understand what is emitted.

Backends - OTLP (Grafana Cloud): managed, simple; push over HTTP. - Prometheus: self-hosted scrape; exposes /metrics. - SQLite: local dev, file-based.

Metrics - Execution counts (success/failure) - Latency histogram - Ratings and health scores

Setup - Select backend via env (ANALYTICS_BACKEND=otlp|prometheus|sqlite). - For OTLP: set endpoint, instance id, token, push interval. - For Prometheus: enable metrics server (host/port) and configure scrape job.

Operational tips - Prefer OTLP/Prometheus in production; SQLite only for local dev. - Watch cardinality; keep metric labels bounded. - Use alerts on error rate and latency p95/p99.