BiteAI Healing Recipe Assistant

Problem It Targets

eatwhats addresses a very specific cooking decision: a user has leftover ingredients, does not want to search blindly through recipes, and needs to respect taste constraints; or the user is tired, cold, caring for a child, and wants a low-effort meal that feels acceptable and comforting. It treats pantry items, dietary restrictions, and emotional context as inputs, so the model generates a recipe under the constraints of available ingredients, prohibited flavors, and the current mood, rather than returning only generic dish names.

How It Works and Where It Fits

The skill is organized as a FastAPI + Python backend service, with the core endpoint POST /api/generate-recipe. It is suitable when users express intents such as “I do not know what to eat,” “how can I use these leftovers,” or “the child does not eat spicy food.” Around that endpoint, it exposes GET /api/pantry to list current ingredients, POST /api/pantry to add ingredients, POST /api/pantry/reset to restore the default pantry, and POST /api/pantry/scan to recognize ingredients from images. GET /api/profile and POST /api/profile/evolve are used to read and update taste preferences. Typical output can include the recipe, nutritional analysis, and a warm note, making it useful for chat assistants, household automation, or lightweight lifestyle workflows. It requires DEEPSEEK_API_KEY and works best when there is clear ingredient context and preference constraints, not for broad recommendations like “what restaurants are nearby.”