TabTenderBot

What it collects

Only what a restaurant, bar or cafe publishes about itself on its own website: the business name, street address, business phone number, website address, opening hours, happy hour, booking link, the features your homepage mentions (a rooftop, a patio, a pool table, live music), and the names and prices of items on its menu. Hours are shown with the month we read them and a link to your site, because they change and yours is the live answer. When your site doesn't list hours, we show OpenStreetMap's instead, labelled as theirs, until your site does. We find venues using OpenStreetMap data (© OpenStreetMap contributors, available under the Open Database License) and Overture Maps places data (© Overture Maps Foundation, available under CDLA Permissive 2.0, with Foursquare's contribution under Apache 2.0), then read the venue's own site.

A venue's page shows average prices per category and the item names and prices behind those averages, with a link to the venue's own website and menu. Item descriptions are never stored, so it is a price list rather than a copy of the menu. The one exception is a short weekly offer written under a day's name (“Mondays: $1 oysters”), which is itself a price. Every page is a snapshot: the venue's own site is always the live one. If you would rather we didn't, tell us and we'll stop.

What it never collects

• Information about people: no emails, names, reviews, or photos.
• Anything behind a login, a form, or a paywall.
• Anything from review sites, delivery or ordering apps, booking services, or social networks. It only reads a venue's own website.

How it behaves

• It follows your robots.txt, including rules written for TabTenderBot and any Crawl-delay.
• It waits at least 10 seconds between requests to a site, uses one connection at a time, and reads at most 20 pages per visit.
• It stops at once if a site answers with an access-denied or too-many-requests response, and it never works around a block.
• It skips pages marked noindex or noai, sites that reserve text-and-data-mining rights, and sites whose terms forbid automated collection.
• A menu file that a venue's own page links to, such as a PDF or a picture of the menu, is read even when the file is marked noindex. Some website builders add that mark to every uploaded file, and on a file it only asks search engines not to list it. A file marked noai, or covered by a data-mining reservation, is still skipped.
• It revisits a site about every 90 days to keep prices current.

How to tell it's really us

Every request is signed with Web Bot Auth (HTTP Message Signatures, RFC 9421). Each one carries a Signature-Agent header naming https://tabtender.app, and the public key it was signed with is published at https://tabtender.app/.well-known/http-message-signatures-directory. A request that says it is TabTenderBot but is not signed with that key is not from us. It runs from a home connection, so it has no fixed IP address to check.

How to block it

Add this to your site's robots.txt and it won't come back:

User-agent: TabTenderBot
Disallow: /

Opt out or ask for a change

Own or run a venue and want us to stop crawling your site, take a page down, or fix something on it? Contact us with your website address. We record the opt-out on our side too, so it holds even if your robots.txt changes.