Woolies

Recipe JSON Schema

When saving a recipe, use this format:

{
  "name": "Beef Tacos",
  "serves": 4,
  "tags": ["mexican", "quick", "weeknight"],
  "ingredients": [
    {"item": "beef mince", "qty": "500g", "search_term": "beef mince"},
    {"item": "onion", "qty": "1", "search_term": "brown onion"},
    {"item": "taco shells", "qty": "1 pack", "search_term": "taco shells"},
    {"item": "cheddar cheese", "qty": "200g", "search_term": "cheddar cheese"},
    {"item": "sour cream", "qty": "1 tub", "search_term": "sour cream"},
    {"item": "iceberg lettuce", "qty": "1", "search_term": "iceberg lettuce"},
    {"item": "tomato", "qty": "2", "search_term": "tomato"}
  ]
}

Key points:

• item is the normalised ingredient name (lowercase, singular where sensible)
• search_term is what to search on Woolworths (can differ from item name)
• tags help with clustering and suggestions
• Skip pantry staples (oil, salt, pepper, common dried spices) unless central

Add to Cart

python3 scripts/cart.py '[{"stockcode": 123, "quantity": 1}]'

• Generates a JavaScript snippet the user pastes into the browser console on woolworths.com.au (must be logged in)
• Always ask before generating

Shopping List HTML

python3 scripts/shopping_list.py '<json>'

• Generates a self-contained HTML shopping list and opens it in the browser
• Writes to /tmp/woolies-shopping-list.html (overwritten each time)
• The page is mobile-friendly with checkboxes that persist via localStorage
• Embedded Add to Cart script - when items include stockcode, the page includes a collapsible panel with the cart JS snippet, a "Select All" button, and a "Copy to Clipboard" button. No need to run cart.py separately.
• Input JSON schema:

{
  "title": "Shopping List",
  "recipes": ["Beef Tacos", "Fried Rice"],
  "items": [
    {
      "item": "beef mince",
      "product": "Woolworths Beef Mince 500g",
      "stockcode": 661766,
      "price": 7.00,
      "quantity": 2,
      "is_on_special": true,
      "savings": 3.00,
      "special_label": "2 for $10",
      "used_in": ["Beef Tacos", "Fried Rice"]
    }
  ],
  "total": 45.50,
  "total_savings": 5.00
}

• title: optional, defaults to "Shopping List"
• recipes: optional, shown as pills at the top of the page
• items[].stockcode: optional, links product name to its Woolworths page
• items[].used_in: optional, shows which recipes need each item
• total / total_savings: optional, shown in a sticky footer

Workflows

The skill supports three modes depending on what the user asks for.

Mode 1: Quick Shopping List

Trigger: User provides a plain list like "milk, eggs, bread, chicken"

1. For each item, search Woolworths (5-10 results, parallel)
2. Rank top 3 by unit price, boosting on-special items within 15% of cheapest
3. Exclude out-of-stock items
4. Present comparison table per item with recommendation
5. Show summary with estimated total and savings
6. Offer to add recommended items to cart
7. Generate HTML shopping list - call shopping_list.py with the final picks (include stockcode from search results so products link to Woolworths). Opens in browser automatically.

Mode 2: Recipe Shopping

Trigger: User provides recipe names like "beef tacos, spaghetti bolognese"

1. Check the library first - run recipes.py list then recipes.py show for any matching saved recipes
2. For unknown recipes - generate ingredients from your knowledge. Present the breakdown and ask the user to confirm/adjust
3. Save confirmed recipes - once the user confirms a recipe's ingredients, save it with recipes.py save so it's available next time. Tell the user you're saving it.
4. Merge ingredients across recipes, summing quantities for duplicates. Call recipes.py overlap to highlight what's shared.
5. Search and optimise per Mode 1
6. Note the overlaps - tell the user which shared ingredients save them money/waste (e.g. "onion is used in both tacos and bolognese")
7. Generate HTML shopping list per Mode 1 step 7

Mode 3: Meal Planning

Trigger: User asks to "plan meals", "what should I cook this week", or gives a number like "plan 5 dinners"

1. Load the library - run recipes.py list to see what's saved
2. Suggest recipe combinations that maximise ingredient overlap:
   • If the library has recipes, prioritise combinations from it
   • Fill gaps with new suggestions from your knowledge
   • When suggesting, explain WHY these cluster well (shared ingredients)
3. Present 2-3 meal plan options as groups, showing:
   • The recipes in each group
   • Shared ingredients count
   • Estimated unique ingredients needed
4. User picks a plan (or mixes and matches)
5. Save any new recipes that the user confirms
6. Run overlap analysis with recipes.py overlap on the final selection
7. Search and optimise per Mode 1
8. Present the full plan with:
   • Recipe list with shared ingredient highlights
   • Optimised shopping list
   • Total cost estimate
   • Savings from ingredient clustering vs buying separately
9. Generate HTML shopping list per Mode 1 step 7

Clustering Logic

When suggesting recipe combinations for meal planning, optimise for:

1. Ingredient overlap - recipes that share proteins, vegetables, bases
   • e.g. mince → tacos + bolognese + cottage pie
   • e.g. chicken thighs → stir fry + curry + tray bake
2. Variety - don't suggest 5 mince dishes. Vary proteins and cuisines
3. Practicality - mix quick weeknight meals with slower weekend ones
4. Seasonal sense - suggest warming food in winter, lighter in summer

When comparing combinations, score them by:

• shared_count from the overlap tool (higher = better)
• total_unique_ingredients from the overlap tool (lower = better)
• Variety of protein/cuisine (subjective, use judgement)

Output Format

Per-item comparison:

## Onion (need: 3 across 2 recipes)

| # | Product | Size | Price | Unit Price | Special? |
|---|---------|------|-------|------------|----------|
| 1 | ...     | ...  | $X.XX | $X.XX/kg   | ...      |
| 2 | ...     | ...  | $X.XX | $X.XX/kg   |          |

-> PICK: [Product] - [reason]

Summary:

## Shopping List Summary

| Item | Recommended | Price | Special? |
|------|-------------|-------|----------|
| ...  | ...         | $X.XX | ...      |

**Estimated Total: $XX.XX**
**Savings from specials: $X.XX**

### Ingredient Overlap
- onion: used in Beef Tacos, Spaghetti Bolognese (buy 3)
- tinned tomatoes: used in Bolognese, Chilli (buy 2 cans)

Important Rules

• Always check in_stock and is_available before recommending
• Use cup_price / cup_measure directly for unit price comparison
• Always save confirmed recipes so the library grows over time
• Always check the library first before generating a recipe from scratch
• Do NOT skip the confirmation step for new recipes
• If a search returns nothing useful, suggest alternative search terms
• Note that Woolworths prices change frequently - prices are as-of search time
• The search API does NOT require login
• The cart API requires the user to be logged into woolworths.com.au in browser