How the QA checks reach a person.

Follow a scheduled check to saved screenshots and run history, then to investigation. Select a box to read what it does.

Choose a component by name

Selected checks

Playwright and Python check selected browser journeys, API responses, parser output and AI behaviour.

Read the code ↗

All six components and their source

Selected checks

Playwright and Python check selected browser journeys, API responses, parser output and AI behaviour.

Read the code ↗

Scheduled checks

Smoke is scheduled every 10 minutes; regression hourly and parser/AI every two hours. Each workflow has its own behaviour and alert policy.

Read the code ↗

Screenshots & history

When a check fails, saved screenshots and run history help a person investigate what happened.

Read the code ↗

Monitoring dashboard

The dashboard presents suite state and execution history. Its paused-monitor display is not a claim that the corresponding GitHub workflow is disabled.

Read the code ↗

Human review

A person checks the saved evidence and the test environment before deciding whether the product has a problem.

Read the code ↗

Smoke email policy

A separate branch decides whether a smoke-failure email should be sent. Its UTC working-hour gate and strictly-greater-than-one-hour cooldown do not erase a failed result. Other workflows must be read separately.

Read the code ↗

This overview leaves out individual assertions, credentials, patient data and private artifacts. It does not run clinic tests. Source revision ebe41cbe080f87084eceede27b3ad91976658456.