mcp-bbc-goodfood

BBC Good Food is a British cooking site, the online home of the magazine of the same name. Its recipes are written and tested by its own cooks, and each one gives its ingredients, its method, its preparation and cooking times, its difficulty, the diets it suits, its nutrition per serving and the stars its readers gave it. The site narrows its recipes along axes of its own: a diet, a cuisine, a kind of meal, a difficulty. Part of the collection sits behind a subscription.

This server connects a chat client to that site. You can read the values each axis takes, search the recipes along them, read one recipe with its ingredients rescaled to the number of people at your table, and switch its quantities between metric and US units. It needs no API key and no account.

Version française

---

Install

One-click install

Claude Code

claude mcp add bbc-goodfood -- npx -y mcp-bbc-goodfood

Claude Desktop, Cursor, and any client using the standard config format

{
  "mcpServers": {
    "bbc-goodfood": {
      "command": "npx",
      "args": ["-y", "mcp-bbc-goodfood"]
    }
  }
}

Node 24 or later is required, and no environment variable has to be set.

With Docker

{
  "mcpServers": {
    "bbc-goodfood": {
      "command": "docker",
      "args": ["run", "-i", "--rm", "ghcr.io/smeet666/mcp-bbc-goodfood:1.1.0"]
    }
  }
}

-i keeps stdin open, which is where the protocol travels, and -t is left out because a TTY rewrites the stream. The container needs outbound HTTPS to www.bbcgoodfood.com, and nothing else: no volume, no port, no credential.

Bundle, without npm

Download mcp-bbc-goodfood-1.1.0.mcpb from the latest release and open it. A client that supports MCP bundles installs it on its own, with no npm and no configuration file to edit. The bundle carries its dependencies, so nothing is fetched at install time.

What you can ask

• "Find me a vegetarian curry that takes under 40 minutes."
• "What diets can I filter on?"
• "Read me that recipe for six, in US cups."
• "Which of these are rated four stars or better?"
• "Scale this ingredient list from a magazine by 1.5."

The ordinary path runs list_filters, then search_recipes, then get_recipe on the path a row carries.

Tools

Tool	What it does
list_filters	Reads the values each axis of the site takes.
search_recipes	Finds recipes, narrowed along those axes.
get_recipe	Reads one recipe, rescaled or in other units on request.
scale_ingredients	Rescales any ingredient list, with no request to the site.

Call list_filters before narrowing a search. The site accepts any value on an axis and answers one it does not know with a total of zero, so a guessed spelling comes back as a confident absence instead of a refusal.

list_filters

Reads the axes the site narrows by, and the values each one takes.

Argument	Type	Required	What it does
query	string, 1 to 80 characters	no	Count the values within one search instead of across the whole listing.

In return: filters, one entry per axis carrying name and label in the site's own wording, argument, which names the argument search_recipes takes for it, and options with each value, its label and its count. A count the site published nothing for is null. option_count says how many options are listed here, which is fewer than the site accepts: the values returned are the most frequent ones, and an option absent from the list is still usable. Counts are measured inside a scope, so passing query counts within one search and leaving it out counts across the listing; the two answer different questions.

search_recipes

Searches the recipes, narrowed along the site's own axes and along restrictions this server applies to the rows it read.

Argument	Type	Required	What it does
query	string, 1 to 80 characters	yes	A dish, an ingredient, a technique.
limit	integer, 1 to 30, default 30	no	Rows to serve.
page	integer, 1 to 334, default 1	no	Which page of rows.
sort	relevant, rating, published or quickest, default relevant	no	How the site orders the rows.
diet	string, 1 to 60 characters	no	A value list_filters publishes.
cuisine	string, 1 to 60 characters	no	A value list_filters publishes.
meal_type	string, 1 to 60 characters	no	A value list_filters publishes.
difficulty	string, 1 to 60 characters	no	A value list_filters publishes.
max_total_minutes	integer, 1 to 1440	no	The whole recipe, in minutes.
max_calories	integer, 1 to 10000	no	Calories per serving.
min_servings	integer, 1 to 50	no	At least this many servings.
min_rating	number, 1 to 5	no	At least this many stars.
exclude_premium	boolean	no	Drop the rows behind the site's subscription.

In return: rows carrying id, which get_recipe takes; title; url; image_url; and rating, which is null where the site published none. Alongside come result_count, rows_seen for the rows the site served before anything was set aside, and total_available. A total marked total_is_ceiling sits on the largest number of rows one search will serve, so it states a floor rather than a count. restrictions_lifted names what was set aside when narrowing made the search fail, and premium_dropped counts the subscription rows removed. The site offers no restriction on subscription rows, so exclude_premium removes them after the page arrives: a page then comes back shorter than the limit asked for, and a short page is not the end of the results.

get_recipe

Reads one recipe, rescaled to a number of servings and in the unit system asked for.

Argument	Type	Required	What it does
id	string, 1 to 200 characters	yes	The page's own path, as a search_recipes row carries it.
servings	integer, 1 to 100	no	Rescale the ingredients to this many servings.
unit_system	metric or us	no	The units the quantities are written in.

In return: title, url, premium, yield_text in the site's own wording such as Serves 4 - 6, yield_count, prep_minutes, cook_minutes, total_minutes, difficulty, diets, author, rating, rating_count, description, ingredients, steps, nutrition with nutrition_per naming the serving it describes, and unit_system. A figure the page states nothing for is null. A recipe behind the site's subscription comes back with premium true, no ingredients and no steps: send the reader to its page rather than reconstructing them. Each ingredient carries scaling, reading scaled, rounded or unscaled.

scale_ingredients

Applies the same arithmetic to any list of ingredient lines, with no request to the site.

Argument	Type	Required	What it does
ingredients	array of 1 to 100 strings, 1 to 300 characters	yes	The lines to rescale, as a recipe writes them.
factor	number, 0.001 to 1000	one of two	What to multiply every quantity by.
from_servings	integer, 1 to 100	one of two	How many people the list feeds as written.
to_servings	integer, 1 to 100	one of two	How many people it should feed.

Pass factor, or the from_servings and to_servings pair.

In return: the rescaled lines in the shape get_recipe returns, each with its original, its text, its amount, amount_max and unit, and its scaling.

Rescaling the quantities

A quantity is stated in the unit that suits it, so a line can come back in a different unit from the one the recipe used: 200 g multiplied by twenty reads 4 kg, and 2 g divided by ten reads 200 mg.

How finely an ingredient can be divided depends on what it is. A loaf can be cut in two, in three or in four; an egg cannot be shared out. A quantity landing between the two is rounded, and the rescaled recipe then departs a little from the proportions of the original. The line carries rounded, and its note says what was done.

The figures are this server's arithmetic, so say they were recomputed when you show them. A recipe whose page states no number of servings cannot be put to a number of people, and the answer says so.

Configuration

Every variable is optional. Set them in the env block of your client config.

Variable	Default	What it does
BGF_USER_AGENT	the project identity	Names your application to the site, with an address where a person can be reached.
BGF_MIN_INTERVAL_MS	1500	Gap between two requests, from 1000 to 60000.
BGF_TIMEOUT_MS	20000	Deadline for one request, from 1000 to 120000.
BGF_MAX_RETRIES	3	Attempts after a transient failure, from 0 to 8.
BGF_CACHE_TTL_MS	900000	How long a page stays in memory, from 0 to 86400000.
BGF_CACHE_MAX_ENTRIES	200	Pages held in memory at once, from 1 to 5000.
BGF_MAX_BODY_BYTES	8000000	Largest page this reads, in bytes, from 100000 to 64000000.
BGF_LOG_LEVEL	error	silent, error, info or debug, written to stderr.

A value outside its range falls back to the default, and the reason is written to stderr.

Errors

Every failure carries one of six codes, a message, and where it helps a hint naming the next move.

Code	What happened	What to do
not_found	The site answered, and holds no such recipe.	Check the path with search_recipes.
invalid_input	The arguments were refused before any request went out.	Read the message, which names the argument.
rate_limited	The site asked this client to slow down.	Wait the number of seconds the hint names and call again with the same arguments. The recipe is still there.
parse_failure	The page loaded and the expected content was absent.	Report it at the issue tracker.
network_error	The request did not complete.	Try again shortly.
timeout	The request passed its deadline.	Raise BGF_TIMEOUT_MS, or ask for fewer rows.

As a library

The layer reading the site is published on its own, with its pacing, its cache and its errors, and with no protocol attached.

import { BbcGoodFoodClient } from "mcp-bbc-goodfood/client";

const client = new BbcGoodFoodClient();
const { data, cached } = await client.searchRecipes({ query: "lasagne" });
console.log(data.rows.length, cached);

listFilters, searchRecipes and getRecipe each answer { data, cached }, and throw an error carrying one of the six codes. The floor between two requests holds here as well.

Pacing and attribution

Requests go out one at a time with at least a second and a half between them, and the floor of one second holds however the server is configured. The User-Agent always ends with the project identity and an address where a person can be reached.

Every result carries the address of the page it was read from, and source names the site. The recipes belong to BBC Good Food and to the cooks who wrote them.

This MCP server is an unofficial project, with no affiliation to BBC Good Food.

Privacy

This server collects nothing about you and sends nothing to its author. It runs on your machine, contacts www.bbcgoodfood.com and nothing else, holds its answers in memory while it runs, and writes nothing to disk. PRIVACY.md states what a request carries and which settings change any of it.

Development

npm install
npm run build:fixtures
npm test
npm run check

Tests run against generated fixtures and make no network request. The live suite, npm run test:live, makes one request per route and runs nightly against the site itself.

Contributing

Bugs, questions and ideas belong in the issue tracker. Pull requests are welcome; opening an issue first helps agree on the shape of the change. See CONTRIBUTING.md.

License

MIT, see LICENSE. The recipes belong to BBC Good Food and to their authors.

---

mcp-bbc-goodfood (français)

English version

BBC Good Food est un site de cuisine britannique, la maison en ligne du magazine du même nom. Ses recettes sont écrites et testées par ses propres cuisiniers, et chacune donne ses ingrédients, sa méthode, ses temps de préparation et de cuisson, sa difficulté, les régimes auxquels elle convient, ses valeurs nutritionnelles par portion et les étoiles que ses lecteurs lui ont données. Le site resserre ses recettes selon des axes qui lui sont propres : un régime, une cuisine, un type de repas, une difficulté. Une partie de la collection est réservée aux abonnés.

Ce serveur relie un client de conversation à ce site. On peut lire les valeurs que prend chaque axe, chercher des recettes le long de ces axes, lire une recette avec ses ingrédients adaptés au nombre de convives, et basculer ses quantités entre unités métriques et américaines. Aucune clé d'API, aucun compte.

Installation

Installation en un clic

Claude Code

claude mcp add bbc-goodfood -- npx -y mcp-bbc-goodfood

Claude Desktop, Cursor, et tout client au format de configuration standard

{
  "mcpServers": {
    "bbc-goodfood": {
      "command": "npx",
      "args": ["-y", "mcp-bbc-goodfood"]
    }
  }
}

Node 24 ou plus récent est nécessaire, et aucune variable d'environnement n'est à renseigner.

Avec Docker

{
  "mcpServers": {
    "bbc-goodfood": {
      "command": "docker",
      "args": ["run", "-i", "--rm", "ghcr.io/smeet666/mcp-bbc-goodfood:1.1.0"]
    }
  }
}

-i garde l'entrée standard ouverte, qui est le canal du protocole, et -t est omis parce qu'un TTY réécrit le flux. Le conteneur a besoin d'un accès HTTPS sortant vers www.bbcgoodfood.com, et de rien d'autre : aucun volume, aucun port, aucun identifiant.

Bundle, sans npm

Téléchargez mcp-bbc-goodfood-1.1.0.mcpb depuis la dernière publication et ouvrez-le. Un client qui gère les bundles MCP l'installe seul, sans npm et sans fichier de configuration à modifier. Le bundle emporte ses dépendances, donc rien n'est téléchargé à l'installation.

Ce qu'on peut demander

• « Trouve-moi un curry végétarien qui prend moins de 40 minutes. »
• « Sur quels régimes puis-je filtrer ? »
• « Lis-moi cette recette pour six, en tasses américaines. »
• « Lesquelles sont notées quatre étoiles ou plus ? »
• « Multiplie par 1,5 cette liste d'ingrédients tirée d'un magazine. »

Le chemin ordinaire va de list_filters à search_recipes, puis à get_recipe sur le chemin que porte une ligne.

Les outils

Outil	Ce qu'il fait
list_filters	Lit les valeurs que prend chaque axe du site.
search_recipes	Trouve des recettes, resserrées selon ces axes.
get_recipe	Lit une recette, adaptée ou dans d'autres unités sur demande.
scale_ingredients	Adapte n'importe quelle liste d'ingrédients, sans requête au site.

Appelez list_filters avant de resserrer une recherche. Le site accepte n'importe quelle valeur sur un axe et répond à celle qu'il ne connaît pas par un total de zéro, donc une orthographe devinée revient comme une absence assurée au lieu d'un refus.

list_filters

Lit les axes selon lesquels le site resserre, et les valeurs que chacun prend.

Argument	Type	Requis	Ce qu'il fait
query	chaîne, 1 à 80 caractères	non	Compte les valeurs dans une recherche plutôt que sur toute la liste.

En retour : filters, une entrée par axe portant name et label dans les termes du site, argument, qui nomme l'argument que search_recipes prend pour lui, et options avec chaque value, son label et son count. Un count que le site n'a pas publié vaut null. option_count dit combien d'options sont listées ici, ce qui est moins que ce que le site accepte : les valeurs rendues sont les plus fréquentes, et une option absente de la liste reste utilisable. Les comptes sont mesurés dans une portée, donc passer query compte dans une recherche et l'omettre compte sur toute la liste ; les deux répondent à des questions différentes.

search_recipes

Cherche des recettes, resserrées selon les axes du site et selon des restrictions que ce serveur applique aux lignes qu'il a lues.