Skip to main content

Service Scorecard

Every week, checks each service against your scorecard (on-call, SLO, monitors, runbook and so on). For each failing rule it opens a pull request that fixes it, with SLO targets taken from the service's own history.

How it works​

  1. Runs when: once a week, over every service in the catalog.
  2. Scores each service: on-call routing, SLO coverage, golden-signal monitors, runbook, README, dependency freshness and branch protection.
  3. Fixes, not tickets: one pull request per repository that fixes every failing rule in it, reviewed by the repository's owners.
  4. Sets SLOs from evidence: targets come from 28 days of the service's own history, with the math in the pull request. It never proposes a target the service has not met.
  5. You get: a scorecard for every service and a pull request per gap. Tracking tickets close once the rule passes.

What it needs​

  • GitHub: read repositories, open pull requests
  • PagerDuty: read on-call and escalation policies
  • Datadog: read SLOs and monitors
  • Jira: track failing rules