- Pass H: consumer-mindset branding review of every screen; output = ready-to-paste ChatGPT image
prompts; must lock the house style first (read brand docs + open existing illustrations) so all
generated art matches the shipped artwork.
- ClaudeBrandingReview.md: canonical House Style prompt prefix + palette + negatives; screen-by-screen
audit (every route); 12 illustration prompts (A1-A12) + glyph set + pack-art prompt, all reusing the
house style; flags 'wire existing iOS art into Android' vs new generation.
- Future.md QA: non-art branding ideas (wire iOS illustrations to Android, consistent glyphs, rotate
privacy messages on auth screens).
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
- Pass B: cover the full depth x round-length matrix (Light/Everyday/Deep/All x 5/10/15), not one combo;
short+long, shallow+deep, every answer type.
- Methodology: THINK AS A CONSUMER (approach from many angles); capture works-but-could-be-better /
feature ideas to Future.md '## QA' (kept separate from the ClaudeReport.md bug log).
- New Future.md seeded with 5 grounded QA improvement ideas (inclusive onboarding options, turn-aware
'waiting to play' copy, rate-limit exemption for high-value pushes, suppress redundant results push,
friendlier paywall error state).
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Don't hand back when context fills: harness auto-summarizes + you continue from the committed
run-state + coverage. Can't self-invoke /compact and don't need to. Commit before interruptible
work; session-start ritual recovers stuck sessions. Only true blockers (denied gated action /
macOS) stop the run.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Round-1 calibration: A & D fit as single batches; B/C/E overflowed and got deferred.
Add a batch-sizing table: B=1 game/chunk, C=1 screen-group/chunk, D=~4 sub-areas,
E=3-5 types/chunk, F=1 dimension/chunk. Chunk = largest unit that finishes+commits in one
window; commit + run-state update per chunk.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Program now Part1 Android QA -> Part2 iOS build -> Part3 iOS QA + cross-platform.
iOS = native SwiftUI (iphone/ scaffold, audit stale at v0.2.0). Decisions: full
Tink-compatible E2EE (Android<->iOS decrypt), working-parity build (no App Store).
Hard constraint: iOS build/run/QA needs macOS (not this Linux box) — Linux = author
Swift + refresh audit only; compile/run/QA deferred to a Mac.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Adds Execution-mode directive: run all passes -> fixes -> re-QA continuously to a flawless
round without checking in; fix anything that BLOCKS progress inline (stale data, crash, build
break, broken nav) to keep going; context limits = checkpoint not stop. Only a denied gated
action (prod deploy / admin write / entitlement toggle) may be surfaced, after all other work.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
UI review now verifies each screen opens correctly from ALL its entry points (inbox/Discuss/
notification, Play/notification, paywall from each gate) and that back (system + in-app)
returns correctly with no dead-ends, exit-app surprises, or two-back/duplicate-stack issues.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Pass B now mandates playing each game end-to-end on both devices (start -> every step ->
finish/reveal/results); launch-only = partial. Reflected in playbook, report run-state,
and coverage (full playthroughs owed in Round 2).
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Reusable QA → fix → re-QA plan. Report-only passes with severity labels, then fix
one-at-a-time by severity, then re-QA until flawless. State/resume lives in ClaudeReport.md
+ ClaudeQACoverage.md. Not yet executed.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
null