How to use this sheet
Every patient-facing feature in the bot, broken into individual test steps. For each row: send the
Send message from a real WhatsApp device (or test harness), confirm the bot does
what Expect describes, then mark the result. Progress and notes save to this browser
(localStorage). The Coverage tag shows the automated test that also guards the behavior.
🤖 Auto-verified steps are pre-marked Pass
Steps tagged Auto ✓ are covered by the offline deterministic suites
(bun test src/ · test:flow:all · deep-link/geo/reminder scripts), which
ran green on 2026-06-14 — skip them in manual QA. The remaining tags
mark what still needs a human or a configured environment:
LLM (live Vertex model — test:intent + test:flow:llm ran green locally 2026-06-15),
Needs env (Supabase/Postgres),
Server (apps/server + Postgres),
Manual (device/vision/PDF/adversarial),
No test (gap). Use the 🤖 Auto /
✋ Needs manual filters above.