SLO burn-rate reporting

The SLO burn-rate report workflow runs every 15 minutes and can be started manually. Configure repository variables for the Prometheus API:

STAGING_PROMETHEUS_URL=https://prometheus-staging.example.com
PRODUCTION_PROMETHEUS_URL=https://prometheus.example.com

If authentication is required, store PROMETHEUS_BEARER_TOKEN as a GitHub Actions secret. The report uses the starter's request and error counters to calculate availability error-budget burn for 1-hour and 6-hour windows against the 99.9% starting SLO. Reports are uploaded as workflow artifacts. Missing configuration skips a scheduled environment without creating noise.

Fast burn (1-hour rate at 14.4x) or sustained burn (6-hour rate at 6x) creates one deduplicated [slo-burn-rate] issue per environment. These thresholds are starting points; tune them to the service's traffic and error budget.

Every successful Record deployment run also uploads deployment-annotation.json, a small vendor-neutral event containing the environment, version, image digest, rollback version, and timestamp. Import this JSON into Grafana annotations or an incident system to place release changes on the same timeline as SLO burn.

For longer-term comparisons, use the quarterly reliability trend report.