ingredient-parsing

name	ingredient-parsing
description	Parse free-text freezer/pantry/fridge items into structured ingredient data with core food terms for recipe matching.

Ingredient Parsing Skill

This skill takes a free-text description of freezer, pantry, or refrigerated items the user wants to use up and extracts structured ingredient data for recipe matching.

Input

• raw_text: The user's free-text input, typically comma-separated (e.g., "1lb frozen chicken thighs, fresh cilantro bunch, pesto sauce, 2 cans black beans").

Workflow

1. Split the input into individual items.
2. For each item, extract the core food ingredient by stripping:
   • Quantities and numbers (e.g., "1lb", "2 cans")
   • Units of measurement (e.g., "lb", "bunch", "cans")
   • Storage/state modifiers (e.g., "frozen", "fresh", "canned", "leftover")
   • Filler words (e.g., "of", "a", "the")
3. Identify whether each item contains meat/seafood for recipe search targeting.

Output

Return a JSON array of objects. Each object has:

• raw: The original text for this item, trimmed.
• core_ingredient: The essential food term for recipe matching (e.g., "chicken thighs", "cilantro", "black beans").
• has_meat: Boolean — true if this item contains meat, poultry, or seafood.
• is_main_dish: Boolean — true if this item is a main dish, entree, or primary protein/meal component (e.g. "chicken thighs", "beef", "salmon"). False if it is a sauce, condiment, dip, dressing, herb, side dish, or basic base ingredient (e.g. "Tzatziki sauce", "hummus", "pesto", "sour cream", "cilantro", "romaine lettuce").

Example

Input: "1lb frozen chicken thighs, fresh cilantro bunch, pesto sauce, 2 cans black beans"

Output:

[
  {"raw": "1lb frozen chicken thighs", "core_ingredient": "chicken thighs", "has_meat": true, "is_main_dish": true},
  {"raw": "fresh cilantro bunch", "core_ingredient": "cilantro", "has_meat": false, "is_main_dish": false},
  {"raw": "pesto sauce", "core_ingredient": "pesto", "has_meat": false, "is_main_dish": false}