¿Puedo vibecodear MealWise?

Build a personal MealWise-like recipe generator in an empty repository.
Use Next.js 15, TypeScript, Tailwind CSS, IndexedDB, and the OpenAI Responses API; do not offer alternative stacks.
Make the first run work with npm install and npm run dev.
Add an .env.example containing only OPENAI_API_KEY.
Create one mobile-first page with a natural-language meal request box.
Let the request include ingredients on hand, budget, servings, dietary preferences, exclusions, time and equipment.
Send it to the model with a strict JSON schema.
Render one complete recipe with title, servings, ingredients, estimated ingredient costs, total cost, cost per person, steps and a practical tip.
Clearly label all costs, allergen handling and nutrition-related content as estimates to verify.
Save accepted recipes locally in IndexedDB.
Include a searchable saved-recipe library and deletion controls.
Provide loading, empty, offline, malformed-model-response and retry states.
Validate input and safely render all model output as text.
Keep the API key server-side and never log meal requests or recipes.
Write focused tests for schema parsing and per-person cost calculation.
Create a README with setup, data location, privacy behavior and limitations.
Do not add accounts, billing, analytics, cloud sync, voice input, push notifications or background jobs.
Do not claim live grocery prices, dietary safety, allergy safety, nutritional accuracy or food-safety guarantees.
Finish by running the tests and listing the exact commands used.