Your own recipe databases¶
Pantry Raider can read recipes straight from a recipe website you or your family keep. Add the site once, and its recipes show up next to everything else:
- Recipe search, with Other recipes switched on. Leave the search box empty to browse the whole database.
- The Cook page, where every recipe in the database is matched against what you have in stock, the same way web recipes are.
- Preview, Cook and Save. Open a recipe to read it, cook it without saving, or save a copy into your own library. A saved copy keeps a credit line and a link back to the original.
Your inventory never leaves your box. Pantry Raider downloads the database's recipe list and does the matching locally.
Adding a recipe database¶
- Open Settings, then Recipes, and find Your recipe databases.
- For a database Pantry Raider already knows, such as Deeserved, press its Add button.
- For any other site, paste the address the site gives you (it usually ends in
/api/pantry-raider), add an access key if the site gave you one, and press Test to check it answers, then Add.
Each database has a switch to hide its recipes for a while, and a remove button. Removing a database leaves the recipes you already saved in your library.
A Pi Remote shows the recipe databases its main server has, so add them on the server.
Deeserved¶
Deeserved is a family recipe box of scanned and transcribed recipe cards, and the first site to offer a recipe database. It needs no access key: anyone can read its published recipes.
For developers: serving a recipe database¶
Any website can become a recipe database by answering three read-only requests. The protocol is called the Pantry Raider Recipe Catalog API, version 1. Pantry Raider calls it from the server, never from a browser, so CORS headers are not needed.
The user pastes your base address, for example
https://example.com/api/pantry-raider. Every path below hangs off it.
Access keys¶
Keys are optional. A request without an Authorization header should get
your public recipes. If you hand out keys, Pantry Raider sends the key as
Authorization: Bearer <key> on every request, and a key you refuse should
get a 401 or 403 answer.
GET {base}/catalog¶
Who you are. Pantry Raider calls this when the user presses Test or Add, and
refuses any address whose answer lacks "api": "pantry-raider-catalog".
{
"api": "pantry-raider-catalog",
"version": 1,
"name": "Deeserved",
"description": "Dee's recipe box",
"url": "https://deeserved.com",
"recipe_count": 58,
"updated_at": "2026-11-07T14:02:11Z"
}
name becomes the database's name in Pantry Raider unless the user types
their own.
GET {base}/recipes?q=&limit=&offset=¶
A page of recipe summaries. limit defaults to 50 and may be capped at 200,
offset defaults to 0, and an empty q lists everything. Sort however suits
you, but keep the order stable so pages do not overlap.
{
"total": 58,
"offset": 0,
"limit": 50,
"recipes": [
{
"id": "87",
"slug": "zuppa-toscana",
"name": "Zuppa Toscana",
"description": "Sausage, kale and sun-dried tomatoes in a creamy broth.",
"image": "https://example.com/media/zuppa.webp",
"url": "https://example.com/recipes/zuppa-toscana",
"servings": "4 to 6",
"total_time": "About 3 hours",
"categories": ["Soups and salads"],
"cuisine": "Italian",
"author": "Liz",
"updated_at": "2026-10-30T18:22:00Z",
"ingredients": ["1 lb spicy sausage", "4 cups chopped kale"]
}
]
}
Put the ingredient lines in the summary. Pantry Raider matches every recipe against the user's stock from this listing alone, so it reads your whole database in a few requests instead of one per recipe. It reads at most 2,000 recipes and keeps the listing for ten minutes.
GET {base}/recipes/{id}¶
One full recipe: the summary fields plus the method.
{
"id": "87",
"name": "Zuppa Toscana",
"ingredients": ["1 lb spicy sausage", "4 cups chopped kale", "6 cups broth"],
"ingredient_groups": [
{"heading": null, "items": ["1 lb spicy sausage", "4 cups chopped kale"]},
{"heading": "For the broth", "items": ["6 cups broth"]}
],
"instructions": ["Brown the sausage.", "Combine everything in the slow cooker."],
"instruction_groups": [
{"heading": null, "items": ["Brown the sausage.", "Combine everything in the slow cooker."]}
],
"notes": ["Works on the stove top in about 40 minutes."]
}
An unknown id should get a 404. Group headings are kept as section headings
when the user saves the recipe; without groups, the flat ingredients and
instructions lists are used.
Field rules¶
| Field | Rule |
|---|---|
id |
Required, a string, stable for the life of the recipe |
name |
Required |
image, url |
Absolute http(s) addresses or null. Images must load without a key |
ingredients, instructions, notes |
Lists of plain-text lines, in order |
categories |
Display labels; they also feed the Cook page's cuisine filter |
total_time, servings |
Free text, shown as is |
| Everything else | Optional. Send null, an empty string or an empty list rather than leaving a field out |
Recipes without an id or a name are skipped. Pantry Raider never writes to
a recipe database.