Flutter AppStore Doc + UI Kit

---

name: flutter-appstore-doc-ui-kit

description: Generate a launch-ready app package for App Store submission: a complete Markdown feature document plus AI-generated Apple-style UI design images for each page and a square-corner app icon. Use when the user wants a Flutter app concept/spec constrained to FVM Flutter 3.35.1, offline-first (no backend), camera + photo library permissions, anti-saturation positioning, and no TODO/temp placeholders.

---

# Flutter AppStore Doc + UI Kit

Generate an App Store-ready pack in gated stages with mandatory user review between stages:

Workflow (Approval-Gated)

1. Generate APP feature docs in Chinese + English and output them.

2. Ask user to download/review docs; wait for: **continue** or **revise docs**.

3. Proceed only after user explicitly approves docs.

4. Confirm image-generation capability:

- If API/model access exists, generate page-level UI images from approved docs.

- If no API/model access, output one unified style prompt package for user-side generation.

5. Ask user to download/review UI images.

6. Proceed only after user explicitly approves UI images.

7. Confirm icon-generation capability:

- If API/model access exists, generate icon via model.

- If no API/model access, generate icon with programmatic SVG->PNG pipeline.

8. Ask user to review icon.

9. Finish only after user explicitly approves icon.

Do not auto-skip any gate. If user asks for changes, revise only that stage and re-submit for approval.

1) Required Inputs

Collect/confirm:

Hard constraints to enforce:

2) Stage 1 — Generate and Deliver Feature Docs (ZH + EN)

Run once:

python3 scripts/generate_appstore_pack.py \
  --app-name "SnapSort" \
  --out ./out/snapsort \
  --locales "en,zh-Hans" \
  --primary-color "#0A84FF"

Then prepare doc outputs first and deliver only docs for review:

If the generator only creates one base spec file, split/translate into two locale files before sending to user.

3) Stage 2 — Generate and Deliver UI Design Images

After user approves both docs, check capability first.

Option A: API/model access available

Use `scripts/generate_ui_ai.py`.

Required auth:

Example:

python3 scripts/generate_ui_ai.py \
  --doc ./out/snapsort/docs/app-feature-spec.en-US.md \
  --out ./out/snapsort/ui \
  --primary-color "#0A84FF"

Outputs:

Option B: no API/model access

Output a single unified prompt package (global style + 8 page prompts + naming rules) for the user to generate images manually in external tools.

Then collect user-generated images and continue to UI review gate.

Do not deliver icon in this stage.

4) Stage 3 — Generate and Deliver App Icon

After user approves UI images, check capability first.

Option A: API/model access available

Generate icon with image model and export PNG.

Option B: no API/model access

Generate icon with deterministic programmatic pipeline (SVG source + PNG output).

Deliver icon artifacts:

5) Quality Validation

Before each approval gate, validate that stage artifacts are complete:

Doc stage:

UI stage:

Icon stage:

6) App Store Safety Guidance

Use `references/review-safety-checklist.md` and keep copy conservative:

Notes