recipe companion

The problem

Finding a recipe that matches exactly what is in your pantry while avoiding specific allergens is tedious. Web searches often return cluttered SEO-heavy pages rather than clean, structured cooking data.

What it does

• Searches a local database of 200 international recipes across 12 cuisines for instant, offline answers.
• Ranks cooking ideas based on ingredients you already have while strictly filtering out allergens and excluded items.
• Generates weekly meal plans that rotate cuisines to ensure variety and creates consolidated shopping lists.
• Constructs traditional Hong Kong style set menus including multiple dishes and a soup.
• Performs provider-neutral web research to pull and normalize recipes from credible online sources when local data is insufficient.

Frameworks & tools

Python 3, TheMealDB API, and local JSON database storage.

Why this beats prompting it yourself

Generic prompts often hallucinate cooking times or suggest repetitive meals. This skill uses a structured search engine with a persistent database, hard-coded dietary safety rules, and a logic-driven meal planner that prevents duplicate cuisines in the same week.

Use cases

• Clear out the fridge by finding recipes that use up specific leftover proteins and vegetables.
• Generate a 7-day dinner plan and a formatted shopping list for a family of four.
• Request a traditional Cantonese dinner set with the "two dishes, one soup" format.
• Sanitize and format web recipes into clean Markdown without the blog fluff.

Known limitations

Dietary tags are primarily focused on vegetarian options. Allergen filtering relies on text matching and is not a medical guarantee.