Skip to main content

PR Risk Review

Featured

Reviews each pull request against the service's tier, incident history, SLO and production traffic, scores its risk from 0 to 100, requests the right reviewers, and checks migrations against real table size.

How it works​

  1. Runs when: a pull request against main is opened or updated.
  2. Reads production first: the service's tier and owner, incidents in the last 90 days on the changed path, the SLO and error budget, peak traffic, and open errors in the touched files.
  3. Checks migrations: against the table's live row count, write rate and the lock the statement takes.
  4. Scores the risk: from 0 to 100, with its components, and ranks findings as blocking, high or advisory, each tied to the production fact behind it.
  5. You get: review comments and the right reviewers requested. At a score of 70 or more it recommends a canary. It never approves, merges or pushes.

What it needs​

  • GitHub: read the pull request, post comments, request reviewers
  • PagerDuty, Datadog and Sentry: read incident, SLO and error history
  • Slack (optional): post blocking migration findings