Offline weekly meal planner and shopping-list generator for home cooks who keep their own recipes and want the week's shopping done from what they actually own.
Pitch: Home cooks who plan meals by hand use bloated recipe platforms with accounts, subscriptions, and imported junk; Meal Queue keeps your own recipes locally, lets you lay out a week on phone, tablet, or Windows, and turns that plan into one merged, scaled shopping list that nudges you to use up what is already in the pantry and freezer first - no accounts, no cloud.
Meal-planning apps mostly solve the wrong problem. They push recipe discovery, ads, and subscription paywalls, then ask you to log into them. The actual weekly chore is narrower: pick meals you already know your household eats, notice what you already have, and produce a shopping list that is not missing half an ingredient and does not tell you to buy garlic again. That loop works fine on a phone in the shop, better on a tablet at the counter, and best on a desktop when you are dragging a month of dinners together. It also should not require an account, a tracker, or a network connection.
- Households that cook from their own recurring recipes and rotate them weekly.
- People who shop from a phone and want one merged list, not one list per recipe.
- Anyone tracking freezer and pantry stock who wants "use this first" prompts instead of duplicate purchases.
- Users who refuse cloud-only food apps and want portable, exportable data.
- Sunday plan. Open the planner on the tablet, drag last week's dinners into the week grid, add two leftovers slots, adjust servings from 2 to 5.
- One list. Tap Make list: ingredients merge across meals by item and unit, scale with servings, and sort by aisle group you defined once.
- Shop. On the phone at the store, tick items off; the list stays checked and survives the app being killed mid-aisle.
- Eat and record. Mark a meal cooked from the plan; the meal gains a "cooked N times, last on " history line for future rotation.
- Use-first nudge. Half a bag of frozen peas is logged in the freezer; the plan shows a badge on meals that would consume it, before you buy more.
- Add your recipes once (name, ingredients with quantity + unit, free-form steps, tags, optional photo). No import service, no scraping.
- Log pantry/freezer stock loosely - item, quantity, unit, optional location and date. Optional, but it unlocks the use-first nudges and duplicate warnings.
- Build a week: pick recipes into day slots (breakfast/lunch/dinner), set servings per meal, add "leftovers" or "eat out" slots.
- Generate the shopping list; review the merged lines, edit or add extras, then shop with checkboxes.
- Mark cooked meals. Repeat next week - last week's plan is a one-tap starting template.
- Local recipe library: title, ingredients (quantity/unit/item), steps, tags, photo, servings. Search and tag filter.
- Week planner with per-meal recipe assignment, servings override, leftovers and eat-out slots, copy-previous-week.
- Shopping list: ingredient merge by item + unit, serving-scale math, manual extra lines, aisle grouping, checkboxes with durable check state.
- Pantry/freezer stock log with optional expiry/freezer dates; use-first badges on meals and "already have" markers on list lines.
- Cook history per recipe (count, last cooked) and simple week archive.
- Adaptive layout for phone, tablet, and desktop windows; touch plus full keyboard/mouse control on Windows.
- JSON export/import of all data (recipes, plans, lists, stock, history).
- Platform-native look per OS (see UX strategy in PLAN.md).
- No recipe import from websites, no barcode/OCR scanning, no photo recipe OCR.
- No nutrition, calorie, allergen, or diet analysis.
- No health, medical, or dietary-advice claims; this is a planning tool, not medical or nutrition guidance.
- No accounts, cloud sync, collaboration, or shared household servers.
- No grocery ordering, price tracking, store APIs, or payment integration.
- No recommendation engine, no AI-generated recipes, no ads or trackers.
- No network requests. The app performs none; there is no analytics or ad SDK.
- All data lives in a local SQLite database inside app storage. Backups are user-initiated file exports (JSON), restored only by explicit user import.
- Permissions: photo/camera access only if the user adds a recipe photo (and only on mobile where a picker is required); file save/open for export and import via the platform picker. No location, contacts, camera roll browsing, notifications beyond optional local plan reminders, no background activity.
- Recipe photos are stored in app storage, never uploaded, and deleted with the recipe and on export opt-out.
- Uninstall or delete-data removes everything; export first if you want to keep it.
Flutter skeleton in place (windows/android/ios, empty adaptive shell, CI). No features yet; the milestone list and backlog live in PLAN.md and the issue tracker.
- M0 Scaffold (done): Flutter project (windows/android/ios), CI, toolchain pins.
- M1 Domain: local database, repository/stock/history entities + tests.
- M2 Recipes: library CRUD, search, photos.
- M3 Planner: week grid, slots, servings, copy-week.
- M4 List: merge/scale/aisle/checkboxes, stock interplay.
- M5 Polish: adaptive per-OS UX verification, accessibility, backup/restore.
- M6 Release: Windows build on a
v*GitHub release.
flutter --version # pinned 3.47.7 stable, see toolchain.json
flutter pub get
flutter test
flutter run -d windows
flutter build windows --release
flutter build apk --debugHost-side Linux checks run inside Docker (dart:stable for analyzer/unit-test
smoke checks; a pinned Flutter image where the Windows lanes cannot substitute).
See PLAN.md for CI layout and packaging.
Android applicationId, iOS bundle identifier, and Windows/MSIX identity derive
from com.infinityball.mealqueue (App Store Connect registration: CREATED com.infinityball.mealqueue, 2026-10-24).
MIT - see LICENSE.