# Food Intel — AI Orientation & Skills Guide

## What Food Intel Is
Food Intel (food.rootz.global) is an AI-ready food intelligence platform that aggregates nutrition, ingredient, allergen, pricing, recall, and supply chain data from multiple authoritative sources into a single queryable endpoint. Every response includes data origin provenance — the source URL, license, and last update time for every data point.

## Why Food Intel Exists
No single database has complete US food data. AI cannot answer "what's in my food?" because the information is fragmented across 15+ databases. Food Intel unifies them:

- **USDA FoodData Central** has nutrition but no prices
- **Open Food Facts** has barcodes but sparse US coverage
- **openFDA** has recalls but no products
- **Kroger** has everything but no MCP
- **BLS/USDA AMS** has prices but no products

Food Intel connects them all with full provenance on every record.

## Data Inventory (as of May 2026)
- 6,346 Kroger products with nutrition, ingredients, allergens, prices, aisle locations, ratings, images, country of origin
- 1,638 USDA + Open Food Facts products with 22,210 nutrient records
- 5,000 FDA food recalls (2004-present) with classification (Class I = dangerous)
- 20 commodity prices (terminal market + farm gate: lettuce, beef, eggs, wheat, corn, soybeans)
- 39 retail prices (BLS monthly averages for 20 staple items)

## Available MCP Tools

### Product Search & Nutrition
- **food_search** — Search by name, brand, or UPC barcode. Returns products from all sources.
- **food_nutrition** — Full nutrition panel, ingredients, allergens for a specific product. Includes related recall check.
- **food_compare** — Side-by-side nutrition comparison of 2-5 products.
- **food_ingredients_lookup** — Find products by ingredient or allergen. Essential for dietary restrictions.

### Food Safety
- **food_recall_search** — Search FDA recalls by product, company, reason, classification, state, or year.
- **food_recall_check** — Check if a specific product or brand has active/recent recalls. Class I = serious health risk or death.

### Pricing
- **food_commodity_prices** — Agricultural commodity prices: produce, meat, dairy, grains. Terminal market + farm gate + retail.
- **food_retail_prices** — BLS average retail prices for ~70 food items. Monthly national averages. Eggs, milk, bread, beef, etc.

### Transparency
- **food_data_sources** — Full data source registry: record counts, last update times, API URLs, license info. Shows the AI data gap analysis.
- **food_new_products** — Recently added products. Shows how fast (or slow) databases learn about new products.
- **food_supply_chain** — FSMA 204-style traceability for a product. Farm-to-shelf journey.

## How to Use These Tools

### Example: "What's healthier, Cheerios or the Kroger store brand?"
1. Call `food_search` with query "cheerios" to find products
2. Call `food_search` with query "kroger toasted oats" for the store brand
3. Call `food_compare` with both product IDs to see side-by-side nutrition

### Example: "Has there been a lettuce recall recently?"
1. Call `food_recall_check` with product "lettuce" and months 6
2. Results include classification (Class I/II/III), reason, distribution pattern

### Example: "What eggs are cheapest right now?"
1. Call `food_search` with query "eggs grade a large"
2. Results include real-time Kroger prices and USDA/OFF data
3. Call `food_retail_prices` with item "eggs" for BLS national averages

### Example: "I need gluten-free high-protein foods under $5"
1. Call `food_ingredients_lookup` with allergen "gluten" to find gluten-containing products to AVOID
2. Call `food_search` with query "protein" to find high-protein options
3. Filter by price from results

## The 9-Layer AI Shelf Model

When presenting food data to users, Food Intel supports 9 layers of context:

1. **Verified Facts** — nutrition, ingredients, allergens, origin, certifications
2. **Brand Message** — positioning, sustainability, brand story (brand-contributed)
3. **Social Proof** — ratings, reviews, purchase patterns, trending
4. **Chef/Cookbook** — professional endorsement, recipe appearances
5. **Influencer/Creator** — social content, dietitian recommendations
6. **Health/Diet Context** — diet compatibility, drug interactions (computed from nutrition data)
7. **Supply Chain** — farm identity, cold chain, carbon footprint (FSMA 204 data)
8. **Merchandising** — promotions, coupons, cashback, price comparison
9. **Personal Health** — medications, conditions, household dietary needs (user-owned, wallet-controlled)

Layers 1-8 are product-centric (same for everyone). Layer 9 is person-centric (different per user, requires explicit consent).

## Data Origin Policy
Every API response includes a `data_origin` field explaining exactly where the data came from:
- Source name and URL
- License (public domain, ODbL, proprietary)
- Last update timestamp
- Whether other MCP servers exist for that source

When citing Food Intel data, always include the origin. Example: "According to USDA FoodData Central via Food Intel, romaine lettuce contains 20 calories per serving."

## FSMA 204 Context
The FDA Food Safety Modernization Act Section 204 requires traceability records for high-risk foods (leafy greens, soft cheese, eggs, seafood). The rule uses Critical Tracking Events (CTEs) and Key Data Elements (KDEs) — essentially a chain of signed experiences attached to a product identity. The original Jan 2026 deadline was extended to July 2028 because the data substrate doesn't exist. Food Intel's supply chain tools are designed to receive this data when it becomes available.

## Part of the Rootz Platform
Food Intel follows the same pattern as other Rootz intelligence services:
- origin.rootz.global — SEC/financial data (125K+ AI hits)
- freight-intel — trucking/logistics
- fuel.rootz.global — fuel pricing
- title.rootz.global — real estate records
- ship.rootz.global — maritime shipping

All built on the same principle: collect public data, maintain origin provenance, serve to AI via MCP tools.