# Recipe Manager — Product Roadmap

**Last Updated:** 2026-03-23  
**Approach:** Agent-driven incremental releases

**Planning terminology:** This file is the strategic **Roadmap**. Active coordinated delivery work belongs in `.harness/*-execution-board.md` (**Execution Boards**). Unscheduled ideas stay in **Backlog** sections.

---

## Philosophy

Each release is **shippable and usable** — no partial features. Agents propose milestones, Paul approves, agents build autonomously.

---

## MVP (v0.1) — "It Works!"

**Goal:** Anne and Elizabeth can create, organize, and cook from recipes  
**Timeline:** 48-72 hours of agent work  
**Deliverables:**

### Core Features
- ✅ **Manual recipe entry** — Form with title, ingredients (list), instructions (steps), source URL, notes
- ✅ **Recipe list view** — Cards with title, description, tags
- ✅ **Recipe detail view** — Full recipe with edit/delete buttons
- ✅ **Basic tagging** — Add/remove tags (Dessert, Main Course, Quick, etc.)
- ✅ **Text search** — Filter recipes by title or ingredients
- ✅ **Cook mode** — Checklist UI for ingredients/steps, screen wake lock

### Technical
- ✅ SQLite database with schema migrations
- ✅ REST API (Express + TypeScript)
- ✅ React frontend (Vite + Tailwind)
- ✅ Docker Compose deployment
- ✅ Basic error handling + validation

### Testing
- ✅ Unit tests for API routes
- ✅ Integration tests for database operations
- ✅ Manual smoke test checklist

### Documentation
- ✅ README with setup instructions
- ✅ API documentation (endpoint specs)
- ✅ User guide (how to add/cook recipes)

### Acceptance Criteria
1. Paul can deploy locally and on paje.ca
2. Anne can add a recipe manually
3. Elizabeth can find the recipe and use cook mode
4. No data loss after container restart

---

## v1.0 — "Daily Driver"

**Goal:** Feature parity with CopyMeThat (minus community)  
**Timeline:** 1-2 weeks (incremental agent work)  
**Deliverables:**

### Recipe Import
- **Browser extension** (Chrome/Firefox) to scrape recipes from any site
  - Parse Schema.org Recipe microdata
  - Fallback heuristic parsing for non-standard sites
  - One-click import to local instance
- **Manual URL import** — Paste URL, backend scrapes + parses

### Enhanced Organization
- **Collections** — Group recipes (Holiday, Meal Prep, Favorites)
- **Advanced search** — Filter by tags, collections, cook time, servings
- **Sorting** — By title, date added, last cooked, random

### Cook Mode Improvements
- **Horizontal tablet view** — Landscape-optimized layout
- **Serving size scaling** — Auto-adjust ingredient quantities
- **Timer integration** — Countdown timers for steps (optional)

### Sharing & Export
- **Print styles** — Minimal, card, full-page
- **Email recipe** — Send as plain text or link
- **Export to JSON** — Backup all recipes

### Sync (Optional)
- **Multi-device sync** — SQLite file sync via Syncthing or Dropbox
- **Conflict resolution** — Last-write-wins with merge log

### Testing
- ✅ E2E tests for import + cook mode flows
- ✅ Browser extension tested on top 10 recipe sites

### Acceptance Criteria
1. Anne can import recipes from her favorite sites in <10 seconds
2. Elizabeth can scale a recipe for 8 people
3. Paul can print a recipe for offline use
4. All tests pass in CI

---

## v2.0 — "AI-Enhanced"

**Goal:** Smart features that make recipe management effortless  
**Timeline:** TBD (depends on AI integration complexity)  
**Deliverables:**

### AI Features
- **Ingredient substitutions** — Suggest alternatives based on dietary preferences or availability
- **Meal planning** — Generate weekly menus from recipe collection
- **Shopping list** — Aggregate ingredients, de-duplicate, categorize by aisle
- **Recipe recommendations** — "You haven't cooked this in 3 months, try it tonight!"
- **Nutrition estimates** — Calorie/macro breakdown via USDA database

### Fintrove Integration
- **Recipe cost tracking** — Link ingredients to grocery prices
- **Budget-aware meal plans** — Optimize for cost + nutrition + preferences

### Advanced Features
- **Voice input** — Add ingredients/steps via speech-to-text
- **OCR import** — Scan recipe cards/cookbooks
- **Recipe rating & notes** — Track favorites, add photos of results

### Mobile App (Optional)
- **React Native** or **PWA** for native mobile experience
- **Offline-first** — Sync when online

### Testing
- ✅ AI feature accuracy benchmarks
- ✅ Load testing (1000+ recipes, complex meal plans)

### Acceptance Criteria
1. AI suggests a valid substitution for missing ingredient
2. Meal plan generation completes in <5 seconds
3. Shopping list works offline
4. Fintrove integration tracks recipe costs accurately

---

## Future Ideas (Backlog)

- **Community recipes** — Optional sharing with friends/family
- **Recipe version control** — Track edits over time
- **Collaborative cooking** — Multi-user cook mode (dinner party!)
- **Video tutorials** — Embed YouTube/TikTok links
- **Equipment tracking** — "Recipes you can make with Instant Pot"
- **Seasonal suggestions** — "Fall comfort food recipes"

---

## Agent Notes

### How to Use This Roadmap

1. **Read PROJECT.md** first for vision + constraints
2. **Pick the next uncompleted milestone** (MVP → v1 → v2)
3. **Break milestone into tasks** (1-4 hour chunks)
4. **Implement ONE task at a time** — commit, test, document
5. **Update this roadmap** when milestones complete

### Decision Authority

- **Agents decide:** Implementation details, libraries, file structure
- **Paul approves:** Milestone completion, new features, major architecture changes

### Communication

- **Progress updates:** After each task completion (commit message + summary)
- **Blockers:** If stuck >30 minutes, escalate to Paul with context
- **Proposals:** New features or scope changes → document in PROJECT.md, ask Paul

---

## Milestone Tracking

| Milestone | Status | Started | Completed | Notes |
|-----------|--------|---------|-----------|-------|
| MVP (v0.1) | 📋 Planned | — | — | Awaiting agent kickoff |
| v1.0 | ⏳ Pending | — | — | After MVP validation |
| v2.0 | 💡 Ideas | — | — | AI integration research needed |

---

_This roadmap evolves as we learn. Agents: update it as you work. Paul: review and approve milestones._