Unit 2: The Workbench — record, mock, assert, cover

Time: ~55 minutes • Lessons: 5 • Modality: UI (workbench)

The testing-oriented half of the workbench, all from the browser UI: capture traffic, replay it as a mock, express what "correct" means with Flow assertions, read regression coverage per method, and inspect intercepted traffic. Still UI-only — where a scriptable equivalent exists (bowire mock / bowire test, the reverse-proxy) this unit links to the CLI (Unit 3) or embedded (Unit 4) unit rather than switching modality inline.

From here on we drive the Harbor Control Center domain (Ship, Dock, Crane, Container, PortCall) via the Combined sample from Bowire.Samples — :5101/bowire, one host, every protocol.

Lessons

Lesson	Topic	What You'll Learn
2.1	Record & Replay	Capture calls, export a .bwr, replay it as a mock from the Mocks rail
2.2	Schema-backed mocks	The mock re-emits the source contract; peer-discovery + coverage gaps
2.3	Flow Assertions	The five-kind (kind, target, operator, expected) expectation vocabulary
2.4	Regression Coverage	The four-state per-method chip: recent · stale · failing · uncovered
2.5	Intercept rail — four postures	Captured · Live overrides · Mock servers · Settings — when to reach for which

Why this unit

Discovering and invoking (Unit 1) is the "does it move" loop. This unit is the "does it move correctly, and can I stand in for it" loop — the surface every operator and QA role leans on.

---

Next: → Lesson 2.1: Record & Replay