DESIGN.md

Part 2b — CI & Alerting Design

No code required here — we want your reasoning.

1. Running this suite in GitHub Actions on every PR

Describe the workflow: what triggers it, the job/steps, how you start the API before the tests, and how a test failure fails the build. A YAML sketch is welcome but optional.

# (optional) sketch your workflow here

Notes:

2. Failure summary to Slack (or structured JSON for alerting)

On a failed run, what would you post, and how do you keep it signal (not noise)?

• What goes in the message: (which fields — failing specs? counts? a link to the run? first error? )
• How it's triggered: (only on failure? on flaky retries? on main only?)
• Keeping it useful: (dedup, thresholds, grouping, who gets pinged, etc.)