mcp-wikibooks-cookbook
Explore and scale recipes from the English Wikibooks Cookbook without an API key.
Search recipes by dish name or ingredient (full-text or title search), getting IDs, titles, URLs, and excerpts.
Read a full recipe including ingredients, equipment, steps, tips, times, difficulty, nutrition, and license info.
Rescale ingredients to a target number of servings, with smart rounding for countable items, unit conversions, and flags for unscalable lines. Works on any English ingredient list, even from outside the cookbook.
Browse recipes by cuisine, dish type, or main ingredient (returns a ranked sample).
Read-only: No edits or uploads; attribution support with CC BY-SA 4.0 credit.
Searches, reads, and scales recipes from the English Wikibooks Cookbook, including browsing by cuisine, dish type, or main ingredient.
Click on "Install Server".
Wait a few minutes for the server to deploy. Once ready, it will show a "Started" state.
In the chat, type
@followed by the MCP server name and your instructions, e.g., "@mcp-wikibooks-cookbookfind a recipe for butter chicken and scale it to 4 servings"
That's it! The server will respond to your query, and you can continue using it as needed.
Here is a step-by-step guide with screenshots.
mcp-wikibooks-cookbook
The Cookbook is the recipe collection of the English Wikibooks, a Wikimedia project. Thousands of recipes are written and maintained there in the open, each with its ingredients, the equipment it calls for, its steps, the tips and variations its authors added, and the categories it files itself under: a cuisine, a kind of dish, a main ingredient. The pages are published under Creative Commons Attribution-ShareAlike 4.0.
This server connects a chat client to that collection. You can search the recipes by dish or by ingredient, list them by cuisine, kind of dish or main ingredient, read one with its ingredients rescaled to the number of people at your table, and rescale any ingredient list of your own. It needs no API key and no account.
Install
One-click install
Claude Code
claude mcp add wikibooks-cookbook -- npx -y mcp-wikibooks-cookbookClaude Desktop, Cursor, and any client using the standard config format
{
"mcpServers": {
"wikibooks-cookbook": {
"command": "npx",
"args": ["-y", "mcp-wikibooks-cookbook"]
}
}
}Node 24 or later is required, and no environment variable has to be set.
With Docker
{
"mcpServers": {
"wikibooks-cookbook": {
"command": "docker",
"args": ["run", "-i", "--rm", "ghcr.io/smeet666/mcp-wikibooks-cookbook:2.0.1"]
}
}
}-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
api.wikimedia.org, and nothing else: no volume, no port, no credential.
Bundle, without npm
Download mcp-wikibooks-cookbook-2.0.1.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.
Related MCP server: mcp-cookbook
What you can ask
"Find me a carbonara recipe in the Cookbook."
"Read it for six people instead of four."
"What Thai soups does the collection hold?"
"Which recipes are built on lentils?"
"Scale this ingredient list by 2.5 for me."
The ordinary path runs from a search to a reading: a row carries an id, the
page key, and get_recipe takes that key.
Tools
Tool | What it does |
| Finds recipes by dish or by ingredient. |
| Reads one recipe, rescaled to a number of servings on request. |
| Rescales any ingredient list, with no request to the wiki. |
| Lists recipes by cuisine, kind of dish or main ingredient. |
search_recipes
Searches the Cookbook, either across the whole text of its pages or across their titles alone.
Argument | Type | Required | What it does |
| string, 1 to 300 characters | yes | A dish, such as |
|
| no |
|
| integer, 1 to 50, default | no | Rows to serve. |
In return: rows carrying id, the page key get_recipe takes; title;
url; image_url; description, the short gloss the wiki keeps for a page; and
excerpt, the matching passage with the highlight markup taken off. Each of the
last three is null where the search offered none. total_available is always
null: the search route publishes no total and no paging, so result_count
counts the rows this call returned.
get_recipe
Reads one recipe, and rescales its ingredients when a number of servings is given.
Argument | Type | Required | What it does |
| string, 1 to 300 characters | yes | The page key, such as |
| integer, 1 to 500 | no | Rescale the ingredients to this many servings. |
| integer, 100 to 20000, default | no | How much of the page's prose to serve. |
In return: title, url, ingredients, equipment, steps, tips for
the notes and variations the page publishes, prep_minutes, cook_minutes,
total_minutes, time_text in the page's own wording, time_phases, category
and categories for every category the page files itself under. A figure the
page states nothing for is null. id names the page that was read, which
differs from the key asked for after a redirect, and redirected_from then names
the key that led there. yield says what the recipe was written for and what it
was rescaled to, and 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 wiki.
Argument | Type | Required | What it does |
| array of 1 to 100 strings, up to 300 chars | yes | The lines to rescale. |
| number, above 0 and up to 100 | one of two | The multiplier to apply. |
| number, above 0 and up to 500 | one of two | How many servings the list is written for. |
| number, above 0 and up to 500 | one of two | How many servings are wanted. |
Pass factor, or the from_servings and to_servings pair.
In return: the factor used, the rescaled ingredients in the shape
get_recipe returns, and scaled_count, rounded_count and unscaled_count.
list_recipes
Lists recipes along the categories the Cookbook files them under. The arguments combine, and the answer says which search was built from them.
Argument | Type | Required | What it does |
| string, up to 80 characters | no | A cuisine or a country, such as |
| string, up to 80 characters | no | A kind of dish, such as |
| string, up to 80 characters | no | The ingredient a recipe is built on. |
| integer, 1 to 50, default | no | Rows to serve. |
In return: the rows search_recipes returns, with query naming the search
this server built from the arguments, and each argument echoed back or null
where it was left out.
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.
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.
Licence and attribution
Cookbook pages are published under
Creative Commons Attribution-ShareAlike 4.0.
Repeating a recipe means crediting Wikibooks and the page it came from, and
sharing what is built on it under the same licence. get_recipe returns both
license and the page url, so an answer carries the credit with it.
Configuration
Every variable is optional. Set them in the env block of your client config.
Variable | Default | What it does |
| the project identity | Names your application to Wikimedia, with an address where a person can be reached. |
|
| Gap between two requests, from 500 to 60000. |
|
| Deadline for one request, from 1000 to 120000. |
|
| Attempts after a transient failure, from 0 to 8. |
|
| How long a page stays in memory, from 0 to 86400000. |
|
| Pages held in memory at once, from 1 to 5000. |
|
|
|
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 |
| The wiki answered, and holds no such page. | Check the page key with |
| The arguments were refused before any request went out. | Read the message, which names the argument. |
| Wikimedia 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. |
| The page loaded and the expected content was absent. | Report it at the issue tracker. |
| The request did not complete. | Try again shortly. |
| The request passed its deadline. | Raise |
As a library
The layer reading the wiki is published on its own, with its pacing, its cache and its errors, and with no protocol attached.
import { WikibooksClient } from "mcp-wikibooks-cookbook/client";
const client = new WikibooksClient();
const { data, cached } = await client.getRecipe("Cookbook:Spaghetti_alla_Carbonara");
console.log(data.title, data.ingredients.length, cached);search 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 between them, and the floor
of half a 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.
Reads go to api.wikimedia.org, the host Wikimedia documents and maintains for
developers. Every result carries the address of the page it was read from, and
the licence it is published under.
This MCP server is an unofficial project, with no affiliation to Wikimedia or to the Wikibooks community.
Privacy
This server collects nothing about you and sends nothing to its author. It runs
on your machine, contacts api.wikimedia.org 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 checkTests 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
service 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 for this server's own code, see LICENSE. The Cookbook pages belong to their authors and to the Wikibooks community, under CC BY-SA 4.0.
mcp-wikibooks-cookbook (français)
Le Cookbook est la collection de recettes des Wikibooks anglophones, un projet Wikimedia. Des milliers de recettes y sont écrites et entretenues à découvert, chacune avec ses ingrédients, le matériel qu'elle demande, ses étapes, les conseils et variantes que ses auteurs ont ajoutés, et les catégories sous lesquelles elle se range : une cuisine, un type de plat, un ingrédient principal. Les pages sont publiées sous Creative Commons Attribution-Partage dans les mêmes conditions 4.0.
Ce serveur relie un client de conversation à cette collection. On peut y chercher des recettes par plat ou par ingrédient, les lister par cuisine, type de plat ou ingrédient principal, en lire une avec ses ingrédients adaptés au nombre de convives, et adapter n'importe quelle liste d'ingrédients. Aucune clé d'API, aucun compte.
Installation
Installation en un clic
Claude Code
claude mcp add wikibooks-cookbook -- npx -y mcp-wikibooks-cookbookClaude Desktop, Cursor, et tout client au format de configuration standard
{
"mcpServers": {
"wikibooks-cookbook": {
"command": "npx",
"args": ["-y", "mcp-wikibooks-cookbook"]
}
}
}Node 24 ou plus récent est nécessaire, et aucune variable d'environnement n'est à renseigner.
Avec Docker
{
"mcpServers": {
"wikibooks-cookbook": {
"command": "docker",
"args": ["run", "-i", "--rm", "ghcr.io/smeet666/mcp-wikibooks-cookbook:2.0.1"]
}
}
}-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 api.wikimedia.org, et de rien d'autre : aucun volume, aucun port,
aucun identifiant.
Bundle, sans npm
Téléchargez mcp-wikibooks-cookbook-2.0.1.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 une recette de carbonara dans le Cookbook. »
« Lis-la pour six personnes au lieu de quatre. »
« Quelles soupes thaïes la collection contient-elle ? »
« Quelles recettes sont bâties sur les lentilles ? »
« Multiplie cette liste d'ingrédients par 2,5. »
Le chemin ordinaire va d'une recherche à une lecture : une ligne porte un id,
la clé de page, et get_recipe reprend cette clé.
Les outils
Outil | Ce qu'il fait |
| Trouve des recettes par plat ou par ingrédient. |
| Lit une recette, adaptée à un nombre de parts sur demande. |
| Adapte n'importe quelle liste d'ingrédients, sans requête au wiki. |
| Liste des recettes par cuisine, type de plat ou ingrédient principal. |
search_recipes
Cherche dans le Cookbook, soit dans tout le texte de ses pages, soit dans leurs seuls titres.
Argument | Type | Requis | Ce qu'il fait |
| chaîne, 1 à 300 caractères | oui | Un plat, comme |
|
| non |
|
| entier, 1 à 50, défaut | non | Lignes à servir. |
En retour : des lignes portant id, la clé de page que get_recipe
reprend ; title ; url ; image_url ; description, la courte glose que le
wiki garde pour une page ; et excerpt, le passage correspondant débarrassé de
son balisage de surlignage. Chacun des trois derniers vaut null là où la
recherche n'en a proposé aucun. total_available vaut toujours null : la route
de recherche ne publie ni total ni pagination, donc result_count compte les
lignes rendues par cet appel.
get_recipe
Lit une recette, et adapte ses ingrédients quand un nombre de parts est donné.
Argument | Type | Requis | Ce qu'il fait |
| chaîne, 1 à 300 caractères | oui | La clé de page, comme |
| entier, 1 à 500 | non | Adapte les ingrédients à ce nombre de parts. |
| entier, 100 à 20000, défaut | non | La quantité de prose de la page à servir. |
En retour : title, url, ingredients, equipment, steps, tips pour
les notes et variantes que la page publie, prep_minutes, cook_minutes,
total_minutes, time_text dans les termes de la page, time_phases,
category et categories pour chaque catégorie sous laquelle la page se range.
Un chiffre que la page n'indique pas vaut null. id nomme la page qui a été
lue, qui diffère de la clé demandée après une redirection, et redirected_from
nomme alors la clé qui y a mené. yield dit pour quoi la recette est écrite et
vers quoi elle a été adaptée, et chaque ingrédient porte scaling, valant
scaled, rounded ou unscaled.
scale_ingredients
Applique la même arithmétique à n'importe quelle liste d'ingrédients, sans requête au wiki.
Argument | Type | Requis | Ce qu'il fait |
| tableau de 1 à 100 chaînes, jusqu'à 300 caractères | oui | Les lignes à adapter. |
| nombre, au-delà de 0 jusqu'à 100 | l'un des deux | Le multiplicateur à appliquer. |
| nombre, au-delà de 0 jusqu'à 500 | l'un des deux | Le nombre de parts de la liste d'origine. |
| nombre, au-delà de 0 jusqu'à 500 | l'un des deux | Le nombre de parts voulu. |
Passez factor, ou le couple from_servings et to_servings.
En retour : le factor employé, les ingredients adaptés dans la forme que
rend get_recipe, et scaled_count, rounded_count et unscaled_count.
list_recipes
Liste des recettes selon les catégories sous lesquelles le Cookbook les range. Les arguments se combinent, et la réponse dit quelle recherche en a été bâtie.
Argument | Type | Requis | Ce qu'il fait |
| chaîne, jusqu'à 80 caractères | non | Une cuisine ou un pays, comme |
| chaîne, jusqu'à 80 caractères | non | Un type de plat, comme |
| chaîne, jusqu'à 80 caractères | non | L'ingrédient sur lequel une recette est bâtie. |
| entier, 1 à 50, défaut | non | Lignes à servir. |
En retour : les lignes que rend search_recipes, avec query qui nomme la
recherche bâtie par ce serveur à partir des arguments, et chaque argument redonné
ou null là où il a été omis.
L'adaptation des quantités
Une quantité est exprimée dans l'unité qui lui convient. Après adaptation, une
ligne peut donc apparaître dans une autre unité que celle de la recette : 200 g
multipliés par vingt donnent 4 kg.
La finesse à laquelle un ingrédient se coupe dépend de sa nature. Un pain se
coupe en deux, en trois ou en quatre ; un oeuf ne se partage pas. Une quantité
qui tombe entre les deux est donc arrondie, et la recette adaptée s'écarte alors
un peu des proportions de l'originale. La ligne porte rounded, et sa note dit
ce qui a été fait.
Les chiffres sont l'arithmétique de ce serveur, donc dites qu'ils ont été recalculés quand vous les montrez. Une recette dont la page n'indique aucun nombre de parts ne peut pas être portée à un nombre de convives, et la réponse le dit.
La licence et l'attribution
Les pages du Cookbook sont publiées sous
Creative Commons Attribution-Partage dans les mêmes conditions 4.0.
Reprendre une recette demande de créditer Wikibooks et la page d'où elle vient,
et de partager ce qu'on bâtit dessus sous la même licence. get_recipe rend à la
fois license et l'url de la page, donc une réponse emporte le crédit avec
elle.
Configuration
Chaque variable est facultative. Elles se posent dans le bloc env de la
configuration du client.
Variable | Défaut | Ce qu'elle fait |
| l'identité du projet | Nomme votre application auprès de Wikimedia, avec une adresse où joindre une personne. |
|
| Écart entre deux requêtes, de 500 à 60000. |
|
| Délai d'une requête, de 1000 à 120000. |
|
| Tentatives après un échec passager, de 0 à 8. |
|
| Durée pendant laquelle une page reste en mémoire, de 0 à 86400000. |
|
| Pages gardées en mémoire à la fois, de 1 à 5000. |
|
|
|
Une valeur hors de sa plage retombe sur le défaut, et la raison est écrite sur la sortie d'erreur.
Erreurs
Chaque échec porte un des six codes, un message, et quand cela aide une indication du geste suivant.
Code | Ce qui s'est passé | Que faire |
| Le wiki a répondu, et n'a pas cette page. | Vérifiez la clé de page avec |
| Les arguments ont été refusés avant toute requête. | Lisez le message, qui nomme l'argument. |
| Wikimedia demande à ce client de ralentir. | Attendez les secondes indiquées et rappelez avec les mêmes arguments. La recette est toujours là. |
| La page a chargé et le contenu attendu est absent. | Signalez-le sur le suivi d'incidents. |
| La requête n'a pas abouti. | Réessayez sous peu. |
| La requête a dépassé son délai. | Augmentez |
Comme bibliothèque
La couche qui lit le wiki est publiée seule, avec son rythme, son cache et ses erreurs, sans protocole attaché.
import { WikibooksClient } from "mcp-wikibooks-cookbook/client";
const client = new WikibooksClient();
const { data, cached } = await client.getRecipe("Cookbook:Spaghetti_alla_Carbonara");
console.log(data.title, data.ingredients.length, cached);search et getRecipe répondent chacun { data, cached }, et lèvent une erreur
portant un des six codes. Le plancher entre deux requêtes tient également ici.
Rythme et attribution
Les requêtes partent une à une avec au moins une seconde entre elles, et le
plancher d'une demi-seconde tient quelle que soit la configuration. Le
User-Agent se termine toujours par l'identité du projet et une adresse où
joindre une personne.
Les lectures vont vers api.wikimedia.org, l'hôte que Wikimedia documente et
maintient pour les développeurs. Chaque résultat porte l'adresse de la page d'où
il a été lu, et la licence sous laquelle elle est publiée.
Ce MCP est un projet non officiel, sans affiliation à Wikimedia ni à la communauté Wikibooks.
Confidentialité
Ce serveur ne collecte rien sur vous et n'envoie rien à son auteur. Il tourne sur
votre machine, ne joint que api.wikimedia.org, garde ses réponses en mémoire le temps qu'il
tourne, et n'écrit rien sur le disque. PRIVACY.md dit ce qu'une
requête emporte et quels réglages changent cela.
Développement
npm install
npm run build:fixtures
npm test
npm run checkLes tests s'exécutent sur des fixtures engendrées et n'émettent aucune requête.
La suite en direct, npm run test:live, émet une requête par route et tourne
chaque nuit contre le service lui-même.
Contribuer
Les anomalies, les questions et les idées ont leur place dans le suivi d'incidents. Les propositions de modification sont bienvenues ; ouvrir un ticket d'abord aide à s'accorder sur la forme du changement. Voir CONTRIBUTING.md.
Licence
MIT pour le code de ce serveur, voir LICENSE. Les pages du Cookbook appartiennent à leurs auteurs et à la communauté Wikibooks, sous CC BY-SA 4.0.
Available Tools
4 toolsget_recipeRead a recipeARead-onlyIdempotent
Read one Wikibooks Cookbook page: ingredients, equipment, steps, yield, time, difficulty, category, and the nutrition panel when the page carries one. 'id' is the page key from search_recipes or list_recipes, such as 'Cookbook:Spaghetti_alla_Carbonara'. A plain dish name is accepted and the Cookbook namespace is added. Pass 'servings' to rescale: a countable thing lands on the smallest share a cook can take out of one of it, which is a whole one where half of one cannot be measured out at all, such as an egg or a whole clove, a half where half of one pours, weighs or splits, such as a can or a packet, and a quarter where a knife or the size of the thing goes further, such as an onion, a slice of bread or a jar; a measurement is moved to a smaller unit before rounding so nothing disappears; and anything that cannot be multiplied is flagged instead. Read 'scaling' on each ingredient rather than doing the arithmetic yourself. A page that groups its ingredients by what they are for, such as a cake and its glaze, states that group on every line; a page that lists them flat states null. Two lines can read alike and belong to different parts of the dish. A page offering several versions of the dish states which one each line came from in 'variant'. Those lists replace one another: one of them is used, even where the procedure says to mix all the ingredients. A measurement the page writes as a conversion template is read as the value and the unit the page wrote, and the counterpart the template computes is left out: nothing here is converted between measuring systems. A line asking for as much as is needed and giving an indication in brackets, such as 'water as required (about 1 ½ cups)', states a quantity: the indication is scaled with the recipe and the wording in front of it is left as the page wrote it. A line offering an alternative in brackets, such as '4 eggs (or 8 egg yolks)', has both branches scaled, since a cook takes one of them. The Cookbook keeps recipes and the book's own chapter indexes in one namespace, and both write 'Ingredients' over a bulleted list. A page carrying no recipe box, no recipe banner and no procedure comes back with empty lists and a note: its bullets are links to other pages rather than things to buy. A heading owns what is nested under it and stops at the next heading of its own level, so a page carrying two recipes returns the first of each part and names the rest in a note. A page that only redirects elsewhere is followed to the page it points at, and 'id' and 'redirected_from' say which page was read. Rescaling needs a yield to scale from. A page that states none comes back as published, and says so. A page yielding no quantities is answered with 'factor' 1 and a note: an empty ingredient list was multiplied by nothing. 'author' and 'rating' are always null: the Cookbook is written collectively and carries no reader score. A time the page does not state is null, never zero, and never inferred from the steps. A page stating phases rather than a total, such as a fermentation and a cooking, publishes them in 'time_phases' with 'total_minutes' null: the phases measure different things and are never added. Pages are published under a licence that requires attribution; 'license' and 'url' carry what to credit.
| Name | Required | Description | Default |
|---|---|---|---|
| id | Yes | Page key from a search, such as 'Cookbook:Spaghetti_alla_Carbonara'. | |
| servings | No | Rescale the ingredients to this many servings. | |
| max_description_chars | No |
Output Schema
| Name | Required | Description |
|---|---|---|
| id | Yes | The page that was read, which differs from 'id' after a redirect. |
| url | Yes | |
| tips | Yes | Notes, tips and variations the page publishes. |
| notes | Yes | |
| steps | Yes | |
| title | Yes | |
| yield | Yes | |
| author | Yes | Always null: the Cookbook is written collectively. |
| energy | Yes | Energy per serving as the page states it. |
| rating | Yes | Always null: the Cookbook carries no reader score. |
| source | Yes | |
| license | Yes | |
| category | Yes | |
| equipment | Yes | |
| nutrition | Yes | |
| time_text | Yes | The time in the page's own wording. |
| categories | Yes | Every category the page files itself under. |
| difficulty | Yes | |
| revised_at | Yes | |
| attribution | Yes | |
| description | Yes | |
| ingredients | Yes | |
| time_phases | Yes | Each duration the page states, in the order it states them. |
| cook_minutes | Yes | The phase the page labels as cooking. Null when it labels none. |
| prep_minutes | Yes | The phase the page labels as preparation. Null when it labels none. |
| total_minutes | Yes | The whole dish, stated only where the page states a total: one duration, or a phase the page itself calls the total. Null where the page states phases and no total, since a fermentation, a marinade or a rest is not cooking time and adding them would answer with a figure nobody published. Read 'time_phases' instead. |
| difficulty_max | Yes | The scale 'difficulty' sits on. |
| redirected_from | Yes | Redirect pages walked to reach this one. Empty when the page was reached directly. |
TDQS
Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?
The description discloses an extensive set of behavioral traits: rescaling rules, handling of groups/variants/conversion templates, redirects, pages without recipe boxes, null fields, and licensing. This goes far beyond the annotations (readOnlyHint, idempotentHint, etc.) and leaves little ambiguity about tool behavior.
Agents need to know what a tool does to the world before calling it. Descriptions should go beyond structured annotations to explain consequences.
Is the description appropriately sized, front-loaded, and free of redundancy?
The description is thorough but presented as a single dense paragraph without bullet points or section breaks, making it hard to scan. Each sentence carries information, but the lack of structure reduces conciseness and readability.
Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.
Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?
Given the tool's complexity, the description covers all expected output fields and edge cases, including scaling behavior, redirects, missing data, and attribution. It complements the output schema well, providing context for fields like 'factor', 'time_phases', and 'redirected_from'. This is a nearly complete specification for practical use.
Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.
Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?
The description adds substantial meaning to 'id' (accepts plain dish names, namespace handling) and 'servings' (detailed rescaling rules) beyond the schema. However, 'max_description_chars' is not mentioned in the description nor described in the schema, leaving a gap for one of the three parameters.
Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.
Does the description clearly state what the tool does and how it differs from similar tools?
The description opens with 'Read one Wikibooks Cookbook page' and lists the specific content returned (ingredients, equipment, steps, etc.), making the tool's purpose explicit. It clearly distinguishes itself from sibling tools like search_recipes and list_recipes by focusing on reading a single page.
Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.
Does the description explain when to use this tool, when not to, or what alternatives exist?
The description gives clear usage context: 'id' is the page key from search_recipes or list_recipes, and passing 'servings' triggers rescaling. It also explains when rescaling fails (no yield). However, it does not explicitly name alternatives or provide when-not-to-use guidance beyond this, so it falls short of a 5.
Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.
list_recipesBrowse the CookbookARead-onlyIdempotent
Browse the Wikibooks Cookbook by cuisine, by kind of dish, or by main ingredient. Give at least one of the three; giving several narrows the result. This is built on the Cookbook's search, because the route this server is allowed to use publishes no way to list the members of a category. What comes back is therefore a ranked sample and not the category itself: it is neither complete nor ordered, and its length says nothing about how many recipes the Cookbook holds on the subject. The Cookbook also holds reference pages on ingredients and techniques, and those rank alongside recipes. Open a row with get_recipe before describing it as a recipe. Use search_recipes instead when there is a dish name to look up.
| Name | Required | Description | Default |
|---|---|---|---|
| limit | No | ||
| cuisine | No | A cuisine or a country, such as 'Italian' or 'Thai'. | |
| dish_type | No | A kind of dish, such as 'soup', 'dessert' or 'bread'. | |
| main_ingredient | No | The ingredient a recipe is built on, such as 'chicken' or 'lentils'. |
Output Schema
| Name | Required | Description |
|---|---|---|
| notes | Yes | |
| query | Yes | The search this server built from the arguments. |
| source | Yes | |
| cuisine | Yes | |
| results | Yes | |
| dish_type | Yes | |
| result_count | Yes | |
| main_ingredient | Yes | |
| total_available | Yes | Always null: no route reports how many recipes a category holds. |
TDQS
Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?
The description adds significant behavioral context beyond annotations: it explains that results are 'a ranked sample and not the category itself,' not complete or ordered, and that reference pages may appear alongside recipes. These caveats are crucial for interpreting results and are not provided by the readOnlyHint/openWorldHint annotations. No contradiction with annotations.
Agents need to know what a tool does to the world before calling it. Descriptions should go beyond structured annotations to explain consequences.
Is the description appropriately sized, front-loaded, and free of redundancy?
The description is longer than minimal but every sentence adds value. It front-loads the purpose, then explains limitations, then usage alternatives. The structure is logical and avoids redundancy, though it could be slightly tightened without losing meaning.
Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.
Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?
The description is complete for a browse tool: it covers what it does, how to filter, what the results are (and their limitations), how to handle non-recipe rows, and when to use sibling tools. With an output schema present, it doesn't need to explain return values. The complexity of open-world ranked results is well addressed.
Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.
Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?
The description adds meaning beyond the schema by stating 'Give at least one of the three' and 'giving several narrows the result,' clarifying the relationship between the filter parameters. It also implies that results are rows to be consumed via get_recipe. Schema coverage is 75%, so the description compensates for the undocumented 'limit' parameter by explaining the overall filtering behavior.
Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.
Does the description clearly state what the tool does and how it differs from similar tools?
The description clearly states the tool's purpose: 'Browse the Wikibooks Cookbook by cuisine, by kind of dish, or by main ingredient.' It uses a specific verb ('browse') and resource, and distinguishes itself from siblings by explicitly guiding users to search_recipes for dish names and get_recipe for opening rows.
Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.
Does the description explain when to use this tool, when not to, or what alternatives exist?
The description provides explicit usage guidance: 'Give at least one of the three; giving several narrows the result.' It also specifies when to use alternatives: 'Use search_recipes instead when there is a dish name to look up' and 'Open a row with get_recipe before describing it as a recipe.' This is clear and actionable.
Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.
scale_ingredientsRescale an ingredient listARead-onlyIdempotent
Rescale a list of English ingredient lines. This contacts no website and works on a list from anywhere, including one the user pasted in. Give either 'factor' directly, or 'from_servings' and 'to_servings' and the factor is computed. Grams, millilitres, ounces and pounds are multiplied and rounded to what a scale can show; a quantity that would fall below one is moved to a smaller unit first, so it never rounds away to nothing. Countable things land where a kitchen can follow, and what decides is the content rather than the container: a can, a packet, a sheet of gelatine or a spoon lands on a half, because half of one pours, weighs or cuts. An egg, a yolk, a white, a zest or a whole clove lands on a whole one, since there is no half of it to measure out, which is what stops an answer like '2.4 eggs'. A bottle, a jar, a block and a slice go one step further, to the quarter, as a whole food a knife shares out does. An approximate measure is scaled as a count, so a pinch taken from 6 servings to 25 is 4 pinches; the note gives the everyday equivalence, and the quantity stays in the unit the line used. That covers the gestures, such as a pinch, a dash, a glug or a handful, and anything named after what holds it, such as a capful or a spoonful. A line with no quantity at all comes back untouched and flagged. A bracketed equivalent, as in '450 g (1 pound)', is scaled with the amount it restates rather than left to contradict it, and so is one written after a slash, as in '500 g / 1.1 lb'. A line offering a choice, as in '2 tablespoons butter OR 30 g margarine', has both branches scaled, so whichever one the cook takes carries the same share. Prefer this over doing the arithmetic yourself.
| Name | Required | Description | Default |
|---|---|---|---|
| factor | No | Multiplier to apply. Use this or the from/to pair. | |
| ingredients | Yes | Ingredient lines, for example ['450 g (1 pound) spaghetti', '5 egg yolks', 'Salt']. | |
| to_servings | No | How many servings are wanted. | |
| from_servings | No | How many servings the list is written for. |
Output Schema
| Name | Required | Description |
|---|---|---|
| notes | Yes | |
| factor | Yes | |
| ingredients | Yes | |
| scaled_count | Yes | Lines whose arithmetic came out exact. |
| rounded_count | Yes | Lines whose value was moved to stay usable, not lines that could have been. |
| unscaled_count | Yes | Lines carrying nothing that can be multiplied. |
TDQS
Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?
Discloses a wealth of behavioral traits beyond annotations: no website contact, rounding to scale readability, unit conversion to avoid rounding to zero, and handling of countable vs measurable items. Also covers bracketed equivalents, slash alternatives, OR branches, and flagging no-quantity lines. This far exceeds what the readOnly/idempotent annotations provide.
Agents need to know what a tool does to the world before calling it. Descriptions should go beyond structured annotations to explain consequences.
Is the description appropriately sized, front-loaded, and free of redundancy?
Although detailed, the description is carefully structured: purpose, input options, rounding behavior, special cases, and a closing recommendation. Each sentence provides a distinct, actionable rule with no filler. Length is appropriate for the tool's complexity.
Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.
Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?
Covers base scaling, rounding, unit conversions, countable/uncountable logic, syntactic variations, and no-quantity handling. With an output schema available, it doesn't need to explain return values, yet it even previews the flag for no-quantity lines. Siblings are unrelated to scaling, so no cross-reference is needed.
Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.
Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?
Schema covers all four parameters, but description adds critical semantics: which parameter combination to use (factor vs from_servings/to_servings) and that the factor is computed from the servings pair. It also clarifies that ingredient lines need not come from a recipe database. This is a strong supplement to the schema.
Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.
Does the description clearly state what the tool does and how it differs from similar tools?
Opens with a specific verb+resource: 'Rescale a list of English ingredient lines.' It clearly distinguishes from sibling recipe search/list/get tools, and explicitly says to prefer this over doing arithmetic yourself. The purpose is unmistakable.
Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.
Does the description explain when to use this tool, when not to, or what alternatives exist?
States when to use the tool: works on a list from anywhere, including pasted input. Provides explicit input mode guidance ('Give either factor directly, or from_servings and to_servings') and names the alternative (doing arithmetic yourself). This is explicit when/when-not guidance.
Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.
search_recipesSearch the CookbookARead-onlyIdempotent
Search the Wikibooks Cookbook for a recipe, by dish name or by an ingredient it uses. 'text' searches the whole of each page, which is what finds a dish from an ingredient inside it; 'title' matches the page name only, which is exact and misses a dish the Cookbook names differently. The Cookbook also holds reference pages on ingredients, techniques and cuisines, and those rank alongside recipes: a row is a page, and only get_recipe can say whether it carries an ingredient list. A full-text row can be a page that merely names the dish: when the titles do not say the words searched for, the notes say so, because the Cookbook links to dishes it does not hold. On a title search the same rows are pages the Cookbook files under another name, such as 'Eggplant' for 'aubergine', since that search opens no page. 'total_available' is null because the search route reports no total and offers no second page. Narrow the query rather than asking for more. Every row carries an 'id', which get_recipe takes.
| Name | Required | Description | Default |
|---|---|---|---|
| limit | No | ||
| query | Yes | A dish, such as 'carbonara', or an ingredient, such as 'guanciale'. | |
| search | No | 'text' reads the whole page; 'title' matches the page name only. | text |
Output Schema
| Name | Required | Description |
|---|---|---|
| notes | Yes | |
| query | Yes | |
| source | Yes | |
| results | Yes | |
| result_count | Yes | Rows returned by this call. |
| total_available | Yes | Always null: the search route publishes no total and no paging. |
TDQS
Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?
Beyond the read-only and idempotent annotations, the description discloses that total_available is null, no second page exists, and title searches may return pages under alternate names. This adds behavioral context not available from annotations alone.
Agents need to know what a tool does to the world before calling it. Descriptions should go beyond structured annotations to explain consequences.
Is the description appropriately sized, front-loaded, and free of redundancy?
Although the description is dense, every sentence adds meaningful context. It opens with the core purpose, then elaborates on edge cases and result semantics with zero filler.
Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.
Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?
With an output schema present, the description focuses on behavioral and contextual details: result row meaning, limitation of total_available, and the relationship with get_recipe. It covers edge cases like pages that only mention a dish, making it complete for the complexity level.
Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.
Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?
The schema describes query and search with brief examples, but the description enriches the search parameter by explaining exactly how text vs title affects results and why title may miss differently named dishes. Limit gets no description, but the guidance to narrow the query implies its purpose.
Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.
Does the description clearly state what the tool does and how it differs from similar tools?
The description clearly states the tool searches the Wikibooks Cookbook by dish name or ingredient, using specific search modes. It distinguishes itself from siblings by noting that only get_recipe can confirm ingredient lists and every row's id feeds into get_recipe.
Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.
Does the description explain when to use this tool, when not to, or what alternatives exist?
The description explains when to use text vs title searches, warns that full-text rows may not exactly match, and explicitly advises narrowing the query rather than paginating. It also points to get_recipe for detail, providing clear alternative guidance.
Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.
Tool Schema Changelog
Recent tool additions, removals, and schema changes observed during successful MCP inspections.
2 tool updates
v1.3.0- Changed
get_recipe11 fields changed- added
Output schema / properties / cook_minutes / descriptionAdded value: +"The phase the page labels as cooking. Null when it labels none." - added
Output schema / properties / id / descriptionAdded value: +"The page that was read, which differs from 'id' after a redirect." - added
Output schema / properties / ingredients / items / properties / groupAdded value: +{ + "anyOf": [ + { + "type": "string" + }, + { + "type": "null" + } + ], + "description": "The part of the dish this line is for, such as 'Glaze'. Null when the page names none. Every part is made, so a caller buys all of them." +} - changed
Output schema / properties / ingredients / items / properties / scaling / descriptionPrevious value: -"'scaled' means the arithmetic was exact. 'rounded' means a countable item was moved to a whole or half unit, or a measurement was demoted to a smaller unit to stay usable. 'unscaled' means the line carries nothing that can be multiplied and was left alone."New value: +"'scaled' means the arithmetic was exact. 'rounded' means a countable item was moved to a whole, a half or a quarter, whichever is the smallest share a cook takes out of one of it, or a measurement was demoted to a smaller unit to stay usable. 'unscaled' means the line carries nothing that can be multiplied and was left alone." - added
Output schema / properties / ingredients / items / properties / variantAdded value: +{ + "anyOf": [ + { + "type": "string" + }, + { + "type": "null" + } + ], + "description": "The alternative list this line belongs to, such as 'Variation II'. Null when the line belongs to the recipe itself. Lines carrying different variants replace one another: exactly one of them is used, whatever the procedure says about mixing everything." +} - changed
Output schema / properties / ingredients / items / requiredPrevious value: -[ - "text", - "original", - "scaling", - "amount", - "amount_max", - "unit" -]New value: +[ + "text", + "original", + "scaling", + "amount", + "amount_max", + "unit", + "group", + "variant" +] - added
Output schema / properties / prep_minutes / descriptionAdded value: +"The phase the page labels as preparation. Null when it labels none." - added
Output schema / properties / redirected_fromAdded value: +{ + "description": "Redirect pages walked to reach this one. Empty when the page was reached directly.", + "items": { + "type": "string" + }, + "type": "array" +} - added
Output schema / properties / time_phasesAdded value: +{ + "description": "Each duration the page states, in the order it states them.", + "items": { + "additionalProperties": false, + "properties": { + "label": { + "anyOf": [ + { + "type": "string" + }, + { + "type": "null" + } + ], + "description": "The phase as the page names it, such as 'Fermentation'. Null when it names none." + }, + "minutes": { + "anyOf": [ + { + "type": "number" + }, + { + "type": "null" + } + ], + "description": "The duration in minutes, or the lower bound of a range." + }, + "minutes_max": { + "anyOf": [ + { + "type": "number" + }, + { + "type": "null" + } + ], + "description": "The upper bound where the page gives a range, null where it gives one figure." + }, + "text": { + "description": "The duration in the page's own wording.", + "type": "string" + } + }, + "required": [ + "label", + "text", + "minutes", + "minutes_max" + ], + "type": "object" + }, + "type": "array" +} - added
Output schema / properties / total_minutes / descriptionAdded value: +"The whole dish, stated only where the page states a total: one duration, or a phase the page itself calls the total. Null where the page states phases and no total, since a fermentation, a marinade or a rest is not cooking time and adding them would answer with a figure nobody published. Read 'time_phases' instead." - changed
Output schema / requiredPrevious value: -[ - "id", - "title", - "url", - "yield", - "ingredients", - "equipment", - "steps", - "tips", - "prep_minutes", - "cook_minutes", - "total_minutes", - "time_text", - "category", - "categories", - "difficulty", - "difficulty_max", - "energy", - "author", - "rating", - "nutrition", - "description", - "attribution", - "license", - "revised_at", - "source", - "notes" -]New value: +[ + "id", + "title", + "url", + "redirected_from", + "yield", + "ingredients", + "equipment", + "steps", + "tips", + "prep_minutes", + "cook_minutes", + "total_minutes", + "time_text", + "time_phases", + "category", + "categories", + "difficulty", + "difficulty_max", + "energy", + "author", + "rating", + "nutrition", + "description", + "attribution", + "license", + "revised_at", + "source", + "notes" +]
- Changed
scale_ingredients1 field changed- changed
Output schema / properties / ingredients / items / properties / scaling / descriptionPrevious value: -"'scaled' means the arithmetic was exact. 'rounded' means a countable item was moved to a whole or half unit, or a measurement was demoted to a smaller unit to stay usable. 'unscaled' means the line carries nothing that can be multiplied and was left alone."New value: +"'scaled' means the arithmetic was exact. 'rounded' means a countable item was moved to a whole, a half or a quarter, whichever is the smallest share a cook takes out of one of it, or a measurement was demoted to a smaller unit to stay usable. 'unscaled' means the line carries nothing that can be multiplied and was left alone."
4 tool updates
v1.2.0- Changed
get_recipe1 field changed- added
Input schema / additionalPropertiesAdded value: +false
- Changed
list_recipes1 field changed- added
Input schema / additionalPropertiesAdded value: +false
- Changed
scale_ingredients1 field changed- added
Input schema / additionalPropertiesAdded value: +false
- Changed
search_recipes1 field changed- added
Input schema / additionalPropertiesAdded value: +false
4 tool updates
v1.0.0- First observed
get_recipe - First observed
list_recipes - First observed
scale_ingredients - First observed
search_recipes
TDQS
Scored across 4 tools
search_recipes and list_recipes both use the Cookbook's search, but their purposes are distinct: search_recipes is for looking up a specific dish/ingredient by text or title, while list_recipes is for browsing by cuisine, course, or main ingredient. get_recipe and scale_ingredients are clearly separate. There is slight potential to confuse the two search-based tools, but the descriptions disambiguate them well.
All four tools follow a consistent verb_noun pattern: search_recipes, get_recipe, scale_ingredients, list_recipes. Each name clearly indicates the action (search, get, scale, list) and the object (recipes or ingredients). No naming style conflicts.
Four tools is a well-scoped count for a read-only cookbook server. Each tool covers a necessary part of the workflow: finding recipes, reading recipe details, scaling ingredient lists, and browsing by category. No redundant or missing core tools.
The tool surface covers the full lifecycle of retrieving cookbook content: search and browse to discover pages, get to read full recipe details, and scale to adjust ingredient quantities. The server's stated purpose is access to the Wikibooks Cookbook (read-only), and there are no significant gaps preventing an agent from completing common tasks.
Maintenance
Related MCP Connectors
- SpoonjoyOAuthapp.spoonjoy
Your personal recipe kitchen: save, fork, and cook recipes, build cookbooks, and shopping lists.
Search, save, organize, cook, and share recipes with any AI assistant.
Baking unit conversion (cups/grams/temp/recipe scaling). convert_amount, convert_temperature
Search, read, and edit your Halite recipe collection and shopping lists.
Related MCP Servers
- FlicenseNot gradedqualityDmaintenanceEnables recipe search, storage, and meal planning using TheMealDB API. Provides comprehensive recipe discovery, automatic recipe collection management, and custom meal plan creation with detailed cooking instructions and ingredients.1-
- FlicenseNot gradedqualityDmaintenanceEnables browsing recipes, searching by category, and viewing detailed recipe information from the Gousto cookbook API via natural language.-
- FlicenseNot gradedqualityDmaintenanceEnables AI assistants to search recipes, compose nutritionally balanced meals, optimize weekly meal plans based on macro targets for family members, and generate consolidated grocery lists from a personal recipe database.-
- AlicenseAqualityDmaintenanceEnables AI assistants to search New York Times Cooking recipes, fetch full recipes with ingredients and steps without login, and browse personal saved recipe box with optional authentication.461MIT