nutrition-estimator

name	nutrition-estimator
description	Wraps the nutrition estimation scripts with skill-level guidance. Knows when to run estimates, how to interpret results, and how to handle unclear measurements for Granny Hudson's recipes.
version	1.0.0

Nutrition Estimator — Granny Hudson

Approximate nutrition, honestly presented.

Purpose

Provides nutrition estimates for recipes using USDA-based data. These are approximations suitable for home cooking — not clinical-grade values. The skill knows when/how to run the estimation scripts, how to handle edge cases, and how to present results honestly.

When to Fire

• After recipe transcription or data changes
• On /nutrition command
• When nutrition data is requested
• When building ebook (ebook-builder may request nutrition data)

Scripts

# Estimate all recipes missing nutrition data
python3 scripts/estimate_nutrition.py

# Preview without saving
python3 scripts/estimate_nutrition.py --dry-run

# Specific recipe only
python3 scripts/estimate_nutrition.py --recipe-id RECIPE_ID

# Re-estimate all (even those with existing data)
python3 scripts/estimate_nutrition.py --force

# Collection-specific (if applicable)
python3 scripts/estimate_nutrition.py --collection granny

Workflow

1. Identify Recipes Needing Estimates

Check recipe data for entries missing nutrition information. Prioritize:

• Recipes with complete, verified ingredient lists (no [UNCLEAR] markers)
• Recipes most likely to be viewed (popular categories: desserts, mains)
• Recipes flagged by users

2. Run Estimation

python3 scripts/estimate_nutrition.py --dry-run  # Preview first
python3 scripts/estimate_nutrition.py             # Then commit

3. Review Results

For each estimated recipe, check:

• Calorie range — Does it seem reasonable for this type of dish?
• Sodium — Flag if >800mg per serving (relevant for heart-smart converter)
• Sugar — Flag if >25g per serving (relevant for diabetic converter)
• Serving size — Is the yield/serving assumption reasonable?

4. Handle Edge Cases

[UNCLEAR] Measurements

• If a recipe has [UNCLEAR] markers on key ingredients, skip nutrition estimation
• Note: "Nutrition estimate unavailable — some measurements are unclear in the original"
• Do NOT guess measurements for nutrition calculation

Regional/Unusual Ingredients

• Some family recipes use regional terms or ingredients not in standard USDA databases
• Flag these for manual review rather than substituting
• Common examples: "a good handful", "cook until it looks right", "Grandma's special mix"

Yield Uncertainty

• If the recipe doesn't specify servings, estimate conservatively
• Note the assumption: "Estimated for X servings based on ingredient quantities"

Nutrition Data Format

Per recipe, store:

{
  "nutrition_per_serving": {
    "calories": 350,
    "fat_g": 12,
    "saturated_fat_g": 4,
    "carbohydrates_g": 45,
    "fiber_g": 3,
    "sugar_g": 18,
    "protein_g": 15,
    "sodium_mg": 580
  },
  "servings": 8,
  "nutrition_source": "estimated",
  "nutrition_date": "2026-03-25",
  "nutrition_notes": "Based on standard USDA values"
}

Health-Conscious Flags

Flag recipes for companion tools:

• Diabetic converter (diabetic-converter.js): sugar >25g or carbs >60g per serving
• Heart-smart converter (heart-smart-converter.js): sodium >800mg or sat fat >8g per serving
• High-protein: protein >30g per serving (positive flag)

Integration

• recipe-validation — Run validation after updating nutrition data
• ebook-builder — Include nutrition facts in cookbook pages
• careful-not-clever — Never invent nutrition data. Estimate honestly or skip.
• cognitive-memory — Remember which recipes have been estimated and any issues found

Honesty Standard

These are ESTIMATES. Always present them as such:

• "Estimated nutrition per serving (approximate)"
• Never claim clinical accuracy
• Round to reasonable precision (no "347.2 calories" — just "350 calories")
• If the estimate is uncertain, say so

---

Soli Deo Gloria — Faithful stewardship of these recipes includes honest nutrition information.