@chronicle.app/foodnoms

Read the FoodNoms app's SQLite database with FoodNomsExtractor, then emit one EatAction or DrinkAction per logged food entry with FoodNomsTransformer. input defaults to ~/Library/Containers/com.algebraiclabs.foodnoms/Data/Documents/db.db; reading it needs Full Disk Access. The connection is read-only via built-in node:sqlite.

chronicle extract foodnoms --limit 0

Records are food entries, newest first, joined to their meal slot. since/until are inclusive bounds on the entry date, and limit: 0 means unlimited. FoodNoms stores dates as UTC text; the extractor adds a Z and turns the entry's blob ID into a dashed UUID.

Each action happens at the entry's timestamp and is keyed by the entry's ID without its type. FoodNoms has no food-or-drink flag, so an entry whose name contains a common drink word becomes a DrinkAction, and anything else an EatAction. Keeping the type out of the key lets that guess change without changing the action's identity.

The object is a Meal, named after the food and keyed by the action it was eaten in (action.sourceId), so the entry's ID is the action's alone. Its category is the meal slot: the built-in slots map to Breakfast, Lunch, Dinner and Snacks, a customized slot uses its own name, and an unnamed custom slot has none. The description joins the slot, brand, amount and calories.

The agent is the iCloud account owner from @chronicle.app/apple. An explicit account object (or null) describes exported data without a host account lookup.

Tests use a synthetic FoodNoms database with an explicit account, so they never read the host's FoodNoms data or iCloud account.