The reason this is more useful than asking a model for a recipe is that the nutrition is attached per portion and the tags are a fixed taxonomy — so high protein, dairy free and quick to cook is a filter rather than a hope. Search first to narrow, then fetch the full record; the previews exist so you are not pulling whole recipes to decide.
A hosted recipe service. Records are structured rather than prose: nutrition per portion, ingredients with quantities in two unit systems, allergens, equipment, numbered cooking steps and storage guidance, all retrievable by slug.
- list_categories — the calorie bands recipes are filed under, from low calorie through to bulking friendly
- list_groups — dietary patterns, health constraints, lifestyle tags, meal types, dish families and macro tags
- list_files — recipe records with name, slug, category and a preview, filterable by category and capped by a limit
- get_file — the full record for one slug
- search_docs — keyword search across records, by ingredient, author, meal type, diet tag or constraint
- Per-portion nutrition on every record: calories, protein, carbs, fat, fibre, sugar and sodium
- Ingredients carrying both metric and US quantities, plus allergens and preparation notes
- Allergy notes with substitutions, and storage guidance for fridge, freezer and reheating
- Two bundled skills: finding recipes that match a constraint, and building a meal plan from them
Nothing to install and no account: it is a hosted HTTP endpoint you add to your client's configuration. Recipe bodies come back in a structured text format rather than as JSON fields, so the model reads the record rather than indexing into it.
One command — npx -y mcp-remote https://vectoralix.com/mcp/PrVaxzTB2TxR