Privacy policy¶
Effective date: October 14, 2026
Who we are and how to reach us¶
Pantry Raider and Forager are run by Dan Marafino, an individual doing business as Syracuse 3D Printing, in Syracuse, New York, USA. In this policy, "we" and "us" mean him. Questions, requests about your data, and complaints all go to [email protected].
This policy covers two things:
- The Pantry Raider app you run on your own hardware. We hold no data about you for a self-hosted install. The first part of this page explains what the app keeps on your device and the few things it sends out.
- Forager, the optional hosted service at forager.pantryraider.app. If you create a Forager account, we do hold some data about you, described in Forager accounts and shares and the sections after it. The same Forager part is also published at forager.pantryraider.app/privacy.
The Forager Terms of Service cover paid plans, refunds, and the rules for shared recipes.
The self-hosted app¶
Pantry Raider is self-hosted: your inventory, recipes, photos, and settings live on your own hardware, and the app sends no telemetry, no analytics, and no usage reports to anyone. The few features that do reach the internet are listed below, each with exactly what is sent, where it goes, and how to turn it off.
This page is written the same way as What needs the internet: every claim here matches what the code actually does, including the imperfect parts. Where something is not yet as good as it should be, this page says so instead of papering over it.
What stays on your hardware¶
Everything, unless a section below says otherwise. In particular:
- Your inventory, expiry dates, and shopping list (in Grocy, on your box).
- Your recipes, meal plan, and cooking history.
- Photos you scan, receipts, and camera feeds.
- Your settings, API keys, and passwords.
- Timers, nutrition logs, audit sessions, and everything on the kiosk.
- The Home Assistant integration, Stream Deck, thermometers, fridge sensors, shelf buttons, and Bandit satellites: all of this talks over your own network and never leaves it.
- The Bluetooth status broadcast for battery displays, when you turn it on, is one-way and carries only counts and numbers (expired, expiring, pending, timers, a probe reading). It never contains item names, tokens, or anything personal, by design.
- Diagnostics and support bundles. The debug log is a rotating file on your device, secrets are redacted from the downloadable copy, and nothing is ever uploaded automatically. Sharing a bundle is always a manual act by you.
There are no accounts, sign-ups, or registrations required to run the app.
What leaves your install, and when¶
AI photo and receipt scanning¶
Only when you configure an AI provider. If you choose Google Gemini, OpenAI, or Anthropic, these go to that provider under your own API key: photos and receipts you scan; recipe photos, PDFs and web pages you import; the ingredient lines of each recipe you save or edit in Pantry Raider's own recipe library, which are parsed into quantities and units automatically; recipes you ask it to tidy; and item names when it estimates shelf life or nutrition, cleans up a barcode lookup (the product's public catalog fields) or suggests recipes from what is in stock. Nothing else from your inventory goes with them, and nothing about you.
On Google's free Gemini tier, Google may use what you send to improve its products, and people at Google may read it. Use a paid key if that matters to you.
If you choose Ollama, everything runs on your own hardware and no image or text leaves your network. If you configure nothing, no AI requests are made at all, and AI-only buttons do not appear.
Forager's managed AI is a fourth option, covered in the Forager section below.
Off switch: set the AI provider to Ollama, or leave AI unconfigured.
Barcode lookup¶
Scanning a barcode sends the barcode number, and nothing else, to the community Open Food Facts database to turn it into a product name. The request identifies the app by name in its user agent; it carries no account, install id, or inventory data. There is no toggle for this because it only happens when you scan, and a scanned barcode is useless without a lookup; if you prefer, type item names in by hand and no lookup occurs.
The optional AI enrichment step then sends the returned public product fields to your configured AI provider (see above); with Ollama or no provider, that step stays local or is skipped.
Recipe search from public catalogs¶
External recipe suggestions are on from the start, using TheMealDB (or Spoonacular if you add a key), so opening the Cook page looks some up. Searches send your query text, and suggestions from stock send simplified ingredient names derived from your inventory (for example "chicken thighs"): up to twelve for TheMealDB, plus any diet or time filter you pick on the Cook page for Spoonacular. No quantities, dates, or anything about you goes along. Recipe pictures in the results load in your browser from the catalog's own site.
Off switch: in Settings, Recipes, set External recipe suggestions to Off (my recipes only).
Your own recipe databases¶
A recipe database you add under Settings, Recipes, is only contacted when you search recipes with Other recipes on, open the Cook page, or preview or save one of its recipes. Pantry Raider downloads the database's recipe list and matches it against your stock on your own box, so nothing about your inventory is sent to it. Searches are matched locally too. If you gave the database an access key, that key goes along with each request.
Off switch: switch the database off, or remove it, in Settings, Recipes.
Importing a recipe from a link¶
Pasting a recipe URL makes your install fetch that page, so the site you pasted sees a normal page request from your network. Nothing is sent anywhere else.
Weather¶
The kiosk weather panel sends your configured location text to Open-Meteo (geocoding plus forecast). If Open-Meteo is unavailable it falls back to wttr.in; with no location configured, wttr.in estimates one from your network's public address. Requests happen only while the weather panel or a weather tile is in use, cached for ten minutes.
Off switch: do not use the weather panel or tile; you can also point the app at a self-hosted Open-Meteo instance in Settings.
Community shelf life¶
Two separate switches, documented in full on Community shelf life and privacy:
- Downloading the aggregated shelf-life table (on by default) fetches a small file from Forager once a day. The request is anonymous and sends nothing about you or your pantry.
- Sharing your own date corrections is opt-in and off by default. Each shared point is anonymous by construction: no account, no install id, no address, day-level dates only. The server publishes a product only after at least five different senders agree, telling senders apart by a scrambled tag of their network that changes every day. The app also caps community values so they can never make a fridge or counter date much longer than its own rules, and never longer at all for meat, fish, eggs or dairy.
Community barcode fixes¶
Two separate switches, documented in full on Community barcode fixes:
- Using the list of fixed product names (on by default) fetches a small file from Forager once a day. The request is anonymous and sends nothing about you or your pantry.
- Sending your own fixes is opt-in, off by default, and needs a linked Forager account. Each fix holds the barcode, the name, brand and category you chose, and what the lookup had said. It is stored with your account so bad fixes can be traced and blocked, but your account is never published with it.
Update checks¶
The app checks GitHub for a newer version so it can show the update notice: a plain unauthenticated request to GitHub for the latest release or version number, cached for a few hours. GitHub sees what any website you visit sees, your network's public address and a generic client signature; the request carries no install id or account. Applying an update downloads code and container images from GitHub and its container registry. Cub firmware is fetched from GitHub releases by your own server and cached locally, so the flasher page works without your browser touching GitHub.
A "Check for updates automatically" switch in Settings, Backups & Updates controls the update notice's background check above; it is on by default. "Install updates automatically", also on by default, is separate: while it is on, a Raspberry Pi or a server set up with install.sh asks GitHub for a new signed release every six hours, whatever the first switch says. Turn both off and the app no longer contacts GitHub on its own. Pressing the Check for updates button always runs a live, one-off check regardless of either switch. If your install has no internet route, a check simply fails quietly and everything else works.
On a server install, one more thing checks on its own. The Docker Compose
stack ships a small Watchtower container that keeps a plain server up to date,
because a server has no host bridge to run the one-button update a Raspberry Pi
appliance uses. Watchtower asks the GitHub container registry once a day
whether a newer Pantry Raider image exists, and pulls it when there is one. It
looks only at the Pantry Raider container, never at Grocy, Mealie, or Ollama,
and it leaves any Watchtower you already run on the same machine alone. It is
the maintained ghcr.io/nicholas-fedor/watchtower image, which works with
Docker 24 or newer.
That daily check runs on its own schedule and is not affected by the settings
switches above, which cover the app's own update notice. The registry sees the
same thing GitHub does: your network's public address and a generic client
signature, with no install id or account attached.
To stop it, either stop that one container:
docker compose stop watchtower
or pin a version in your .env so nothing newer is ever picked up:
PANTRYRAIDER_TAG=0.21.0
Use the version you run now, shown in Settings, Backups & Updates. A lower number moves the server back to an older release.
A minor line such as 0.21 works too: the daily check then still runs, but
only ever picks up patch releases on that line.
Raspberry Pi appliances do not run Watchtower at all.
Remote access¶
Off by default. If you enable it, traffic to your kitchen flows through the tunnel you chose: your own Cloudflare Tunnel, or Forager's built-in remote access (covered below). On your LAN, nothing is exposed to the internet.
Backups you push to cloud storage¶
Pushing a backup to your own rclone remote (S3, Drive, and so on) is a manual action that sends the backup archive to the storage account you configured. Secrets are left out unless you explicitly include them. Nothing is ever backed up off your hardware automatically.
Product picks¶
The Shop page links to our product picks on pantryraider.app. Some links on that page are Amazon affiliate links, so a purchase after clicking supports the project at no cost to you. These are plain links your browser follows if you click them; the app itself sends nothing to Amazon or to the website, and no purchase or browsing data ever comes back to it or to us.
Forager accounts and shares¶
Forager is the optional cloud companion. The app is fully usable without it; nothing in this section applies until you create an account and link a kitchen.
What an account stores: your email address, a password hash (never the password), and, if you enable them, two-factor and passkey credentials (a passkey stores only its public half; the secret stays on your device). Every sign-in, pairing, and instance token is stored only as a hash. Forager does not ask for or store your name, address, or phone number.
What a linked kitchen registers: a name you choose, the app version, and the deployment mode, so your account page can show your kitchens. No LAN addresses or user IPs are stored for it.
Payments: handled by Stripe. Forager stores only Stripe's customer and subscription identifiers and the subscription status; card numbers and billing details never touch Forager's servers.
Managed AI: photos and text you scan through Forager's AI pass through its server in memory to the upstream model (Google Gemini) and are never written to disk, database, or logs. What is kept is a usage ledger of token counts per month, which is how your quota is metered.
Recipe shares: a share link is a public web page holding exactly the recipe you shared, nothing else from your library. Links are unguessable, you can revoke any of them at any time from your account, and pages reported by several people are taken down automatically. Share links do not expire on their own; they stay up until you revoke them. Sending a recipe by email sends the recipient's address to the mail provider to deliver that one message; the address is not added to any list. When a visitor who is not signed in reports a share page, the report record keeps a short one-way hash of their network address, never the address itself, so repeat reports from the same person count once (see Recipe reports below). Forager's server does not keep request logs, so page addresses and visitor addresses are not recorded there either.
Community recipes: submitting a recipe to the community publishes its title, ingredients, steps, and your attribution text for other kitchens to browse. Your email is never shown; ratings and reports are tied to accounts internally to prevent abuse.
Recipe photos: when you share a recipe that has a photo, your kitchen uploads the photo to Forager, which keeps its own copy so the recipe still shows its picture when your kitchen is offline. Before the copy is saved it is re-encoded and stripped of camera details such as GPS position. Share pages and the community recipe pages on pantryraider.app load that photo from forager.pantryraider.app, so anyone who opens one of those pages fetches the photo from Forager. If a recipe's photo is a link to another website instead, the viewer's browser loads it from that website. Recipes you upload on the Forager website (typed, as a PDF, or as a photo of a recipe card) are read by Forager's AI provider to turn them into a draft you review; the uploaded file itself is not kept.
Cloud backups (Premium): a backup you push to Forager is stored on Forager's server encrypted in transit but not encrypted at rest; treat it accordingly if your backup includes secrets. Only your account can list or download it, and only the newest three are kept.
Remote access through Forager: your kitchen gets a public
yourname.forager.pantryraider.app address. Be aware of what that means:
the secure connection from your browser ends at Forager's server, which
relays traffic to your kitchen over an encrypted WireGuard link. Forager
does not log or inspect that traffic, and no access logging is configured,
but the traffic does pass through the server, so it is not end-to-end
encrypted between your browser and your kitchen. If that trade-off is not
right for you, use your own Cloudflare Tunnel or a VPN instead.
Bot checks: the signup page uses Cloudflare Turnstile, which means Cloudflare sees the visit the way it does on any Turnstile-protected site.
Email: Forager sends email only for verification, password resets, and recipe shares you initiate, delivered through a transactional mail provider. There are no newsletters or marketing emails.
Deleting your account: you can delete your Forager account yourself, from the bottom of your account page. Deletion asks you to confirm who you are, cancels any paid subscription immediately, and removes your sign-in details, sessions, two-factor material, passkeys, paired kitchens, cloud backups (including the stored files), usage history, and your share links along with their photos. Recipes you published to the community remain, with your name replaced by "a former member". If you prefer, or cannot sign in, [email protected] can do it for you.
Cancelling a subscription: cancel online any time from your account page, in as few steps as subscribing took. Your plan stays active until the end of the period you already paid for and simply does not renew.
Recipe reports: if someone reports a shared recipe without being signed in, the report is identified only by a short one-way hash, never a stored IP address, and report records are deleted after 90 days.
Free trial record: an anonymous per-install id is kept with the record of which install has used its free trial, so one copy of the app gets one trial. That record stays after an account is deleted, holds no email address, and does not identify you.
Where Forager data is kept¶
Forager runs on a virtual private server we rent in the United States. Its database, cloud backups, and recipe photos are stored there. If you use Forager from outside the United States, your data is transferred to and handled in the United States.
Service providers Forager uses¶
Forager relies on a few outside companies to do specific jobs. Each one gets only what that job needs.
| Provider | What it does for Forager | What it receives |
|---|---|---|
| Stripe | Takes subscription payments, sends receipts, runs the billing page | Your email, payment details you enter on Stripe's pages, and your Forager account number so the payment is matched to you |
| Google (Gemini API) | Runs managed AI: photo and receipt scanning, recipe formatting, and the other AI features | The photo, receipt, or text of each request, without your email or name |
| Google (Sign-In) | Optional "Continue with Google" sign-in | Your Google sign-in; Forager gets back only your email address and whether Google has confirmed it |
| Resend | Delivers verification, password reset, and recipe share emails | The recipient address and the message |
| Cloudflare | The bot check on the signup page (Turnstile), and DNS for the pantryraider.app domain | Your visit to the signup page, as on any site that uses Turnstile |
| Our hosting provider | The US server Forager runs on | Everything Forager stores, held on its hardware on our behalf |
We do not sell personal data or share it for advertising. We would disclose data only if the law required it, and we would tell you unless the law forbids that.
Your rights and how to use them¶
Wherever you live, you can ask us to:
- See the data Forager holds about you. Your account page shows your email, plan, kitchens, usage, backups, and share links.
- Correct it. Most of it you can change on your account page; to change your email address, write to support.
- Delete it. Delete your account yourself from the bottom of your account page (see Deleting your account above), or ask support to do it.
- Export it. You can download your cloud backups from the app, and on request we will send you a copy of your account data in a common machine-readable format.
- Object to or limit how we use it, or withdraw consent you gave.
To make a request, email [email protected] from the address on your account, so we know the request is really yours. We answer within 30 days and do not charge for it. You will not be treated differently for using any of these rights. If you are in the EU, the UK, or a US state with its own privacy law, you also have the right to complain to your local data protection authority, though we would like the chance to fix the problem first.
For a self-hosted install there is nothing for us to act on: all of that data is on your own device, and you can view, change, export, or delete it there.
Children¶
Pantry Raider and Forager are not meant for children under 13, and Forager does not knowingly hold data about them. If you believe a child has created a Forager account, write to support and we will delete it.
Community shelf life data¶
Covered in full, in plain language, on Community shelf life and privacy: what a shared point contains, what is never shared, and how the aggregated table is built so no single kitchen's data is ever visible.
What we never collect¶
- No analytics, telemetry, or usage tracking of any kind, in the app or in its documentation pages.
- No advertising, no tracking pixels, no third-party trackers.
- No sale or sharing of data with anyone, ever. There is nothing to sell: for a self-hosted install without Forager, we hold no data about you at all.
- No location data, beyond the town name you type in for the weather panel, which goes only to the weather service.
Data retention on Forager¶
- AI images and prompts: never stored; they exist only for the duration of the request.
- Token usage: kept as monthly totals for quota metering and billing history.
- Recipe shares and community recipes: kept until you revoke or delete them.
- Cloud backups: the newest three per account; older ones are deleted when a new one arrives.
- Anonymous shelf-life observations: kept to build the aggregated table; they contain nothing that can identify you.
- Barcode fixes: kept with your account while it exists, and deleted with it. A fix you send again for the same barcode replaces your earlier one.
- Account records: kept while the account exists, removed when you delete the account (see deleting your account above).
- Billing records: Stripe keeps payment and invoice records for as long as the tax and accounting rules it follows require. Forager's own copy of your subscription status goes when your account does.
- Recipe photos: kept while the recipe or share that uses them exists.
- Housekeeping runs daily: expired sign-in tokens, verification and reset links, pairing codes, and stale sessions are purged automatically, and recipe-report records are deleted after 90 days. Tokens are stored hashed and single-use even before they are swept.
Changes to this policy¶
If we change this policy in a way that matters, we update the effective date at the top and, for changes that affect Forager accounts, email account holders before the change takes effect. Changes are recorded here, newest first.
-
2026-10-14: added who runs the service and how to reach us, where Forager data is kept, the named service providers, your rights and how to use them, recipe photo hosting, and children's data. Nothing about what the app or Forager collects changed.
-
2026-07-15: self-serve account deletion, online cancellation, hashed report identifiers, and daily retention housekeeping shipped; the policy now describes them.
- 2026-07-15: first version of this policy.