docs: record Round 2 QA and the Batch 03 log entry

Round 2 covers every screen Batch 03 built, pass by pass, with each partial and
not-run row carrying why. Three new standing gaps, all of them things nobody has
done rather than things that failed: TalkBack has never been run, text has never
been scaled, and nothing has run at minSdk.

The rule this session keeps demonstrating is now stated plainly in both files:
the defects in this project are found by running it, not by reading it.

docs/history/DEVELOPMENT_LOG.md

@@ -32,6 +32,69 @@ written and stay true. It is exempt from review for the same reason a receipt is
 
 ## Entries
 
+### 2026-08-18 — Batch 03 shipped: every screen the app has, and three defects only looking could find
+
+All six Batch 03 issues closed and the milestone with them. Period stopped being
+a data layer with a debug surface and became something you can actually use:
+onboarding that ends in a forecast, a Today screen whose six states each say
+something different and true, two-tap logging, a calendar, and Insights.
+
+**Placeholder artwork, as §42 asks for it.** Every illustration and calendar
+marker is a Compose vector path behind a replaceable name — no raster anywhere
+under `src/`. The visual language is overlapping circular forms and nothing
+else, because §42's forbidden list is a product decision: this app gets opened
+in public and a glance over a shoulder should learn nothing. Note the deliberate
+contrast with `docs/data/img/`, where a placeholder is still forbidden and #8
+stays open; `docs/design/README.md` now explains why the two differ rather than
+leaving it looking inconsistent.
+
+**Three defects, none findable by reading:**
+
+1. **Dark mode was broken for the whole of Batch 01.** `PeriodTheme` never
+   wrapped its content in a `Surface`, so any `Text` without an explicit colour
+   inherited Material's default — black — and the app's background never
+   painted. Light mode looked right by accident, because dark-on-cream is what
+   was wanted anyway. 129 tests were green throughout.
+2. **"Period ended" appeared to do nothing.** The logic was correct: a period
+   that ends today still includes today, so the state genuinely does not change.
+   The screen was identical afterwards and the button read as broken, which is
+   worse than being broken somewhere visible.
+3. **The today-underline collided with the spotting dot** on the one day that
+   was both. Invisible in the colour screenshot and obvious the moment it was
+   converted to greyscale — which is how §43 says to check, and why the check is
+   written that way rather than trusting that shapes differ because they were
+   designed to.
+
+Two more came from tests doing their job: the typical-range quartiles were
+indexed off `size` rather than `size - 1`, so a regular cycle with one long gap
+was reported as "typical range 29–61 days" on the screen whose only job is to
+say what has been learned; and Room's own executor meant a test's virtual clock
+returned before the writes landed, which read as an empty database rather than
+as a timing bug.
+
+**The habit is now five pieces of work old and worth writing down as a rule: the
+defects in this project are found by running it.** Three guards were green over
+exactly what they claimed to check, dark mode was broken for a whole batch, and
+a button looked broken while behaving correctly. None of it was visible in the
+source.
+
+- **Closed:** #15, #16, #17, #18, #19, #20 — and the `Batch 03 — Core UX`
+  milestone.
+- **Next action:** Batch 04 — Fertility. File its issues first; the milestone is
+  empty. The work is estimated ovulation and the fertile window from the
+  forecast and a configurable luteal-phase assumption (§17, §18), the calendar
+  and Today states that already have their shapes and their `null` placeholders
+  waiting, and the not-contraception disclaimer everywhere fertility appears.
+  `CycleStatus` and `CalendarMarks` both already take fertility parameters and
+  currently receive null — Batch 04 is largely filling those in rather than
+  adding surfaces.
+- **Blockers:** Unchanged. #8 needs the three branding marks drawn by a person.
+  #9 needs the Command Center webhook URL and secret, which are not readable
+  from this machine. Neither blocks Batch 04. Worth doing before much more UI:
+  run TalkBack once, scale the font to maximum once, and run the app once on a
+  device at `minSdk` 26 — all three are recorded as standing gaps and none has
+  ever been done.
+
 ### 2026-08-18 — Batch 02 shipped: the prediction engine, and three bugs the tests found
 
 All five Batch 02 issues closed and the milestone closed with them. The app now

docs/qa/ClaudeQACoverage.md

@@ -13,6 +13,40 @@ Review trigger: Any QA round run
 > like a pass that succeeded, and that is how untested code ships believing it
 > was tested.
 
+## Round 2 — 2026-08-18 at `19edf4c`, partial
+
+Batch 03 built every screen the app has, and each was driven by hand on the
+emulator as it landed rather than in one pass at the end. Same environment:
+`PeriodQA`, API 36, Pixel 6 profile, debug build.
+
+| Pass | Result | Notes |
+| --- | --- | --- |
+| A — First run | **Pass** | Clean install reaches onboarding; all seven screens walked; the flow produces a forecast and a relaunch goes straight to Today. |
+| B — Core loop | **Pass** | Two-tap logging confirmed on the device: "Started period" → "Yes — today" → "Logged ✓". Editing, ending, and spotting reclassification all exercised. |
+| C — Failure paths | **Partial** | Duplicate-date logging, an end date before a start, and a future date in the picker all handled. Airplane mode, denied permissions and a process killed mid-write still untried. |
+| D — Persistence | **Partial** | Onboarding completion, records and settings all survive force-stop and relaunch. Reboot and update-over-install untried. |
+| E — Forecast under hard histories | **Partial** | The §51 cases pass as unit tests against both engines, and the not-yet path was driven through the UI. The variable and outlier histories have still not been entered by hand. |
+| F — Notification privacy | **Not run** | Nothing sends a notification yet — Batch 05. The onboarding *choice* was verified: Discreet is selected before the user touches anything. |
+| G — Accessibility | **Partial** | Every calendar day, the confidence indicator and the hero countdown carry content descriptions, checked by reading the view hierarchy. **The calendar was verified in actual greyscale** and all five marks remain distinguishable. TalkBack itself, and font scaling, still untried. |
+| H — Data ownership | **Partial** | Delete-all covered by an instrumented test. Export, app lock and artifact inspection do not exist yet. |
+
+### What driving it found that tests did not
+
+Three defects, none of which any unit test would have caught:
+
+1. **Dark mode was broken for the whole of Batch 01.** `PeriodTheme` never
+   wrapped its content in a `Surface`, so text without an explicit colour
+   inherited black and the app background never painted. Light mode looked
+   correct by accident.
+2. **"Period ended" appeared to do nothing.** The logic was right — a period
+   ending today still includes today — but the screen was identical afterwards,
+   so the button read as broken.
+3. **Today's underline collided with the spotting dot**, on the one day that was
+   both. Invisible in the colour screenshot, obvious in greyscale.
+
+The pattern is now four rounds old and worth stating as a rule: **the defects in
+this project are found by opening it, not by reading it.**
+
 ## Round 1 — 2026-08-18 at `adc5075`, partial
 
 Not a full round. Batch 01 produced the first build, and this covered the two
@@ -74,6 +108,12 @@ at its own `minSdk`.**
 Things no round has ever covered, carried forward until they are. This list
 existing is not a failure; it not existing while the gaps do is.
 
+- **TalkBack has never actually been run.** Content descriptions are written and
+  were checked by reading the view hierarchy, which proves they exist and not
+  that they make sense in sequence. Pass G is not complete until somebody
+  navigates a screen with their eyes shut.
+- **Font scaling untried.** The hero number is 72sp; at the largest accessibility
+  scale it may not fit beside anything.
 - **No device at `minSdk`.** Everything so far ran on API 36. The two NewApi
   bugs above would have crashed on Android 8 through 13 and nothing except lint
   could see them. Lint is a good guard and it is not a substitute for one run on

docs/qa/ClaudeReport.md

@@ -19,19 +19,22 @@ Review trigger: Any QA round run
 
 ## Current run-state
 
-- **Last QA round:** Round 1 — 2026-08-18, partial (passes A and B; C, D and H partial)
-- **Last verified build SHA:** `adc5075`
+- **Last QA round:** Round 2 — 2026-08-18, partial (A and B pass; C, D, E, G and H partial; F not applicable yet)
+- **Last verified build SHA:** `19edf4c`
 - **Last tested device / environment:** emulator `PeriodQA`, API 36, Pixel 6 profile, debug build
-- **Overall status:** The Batch 01 foundation works end to end on a device — log
-  a period, see a forecast with an honest window and confidence, edit it and
-  watch the forecast move — and the first partial QA round found two defects
-  that 70 passing unit tests could not: tapping the app's primary button twice
-  crashed it, and two date APIs on the recalculation path would have crashed
-  every device below Android 14. Both are fixed, with regression tests. That
-  ratio is the honest summary of where this project stands: the logic is well
-  covered and the product has barely been used, so the next round matters more
-  than the last one. Nothing has been tested at the minimum Android version this
-  app claims to support.
+- **Overall status:** Three batches are done and the app is now a real product
+  you can use end to end: onboarding that ends in a forecast, a Today screen
+  whose six states each say something different and true, two-tap logging, a
+  calendar whose marks survive greyscale, and an Insights screen that mostly
+  declines to make claims it has not earned. The prediction engine is the one
+  §12 specifies and is measurably better than the prototype it replaced. What
+  the rounds keep showing is that this project's defects are found by opening
+  the app rather than by reading it — dark mode was broken for an entire batch
+  with 129 tests green, and the calendar's marks collided in a way only a
+  greyscale screenshot revealed. The gaps that matter now are the ones nothing
+  has touched: TalkBack has never been run, text has never been scaled up, and
+  nothing has run on a device at the minimum Android version this app claims to
+  support.
 
 ## Open defects