Yeetfind

Yeetfind is a web search engine that keeps nothing about the people who use it. It searches a short list of documentation sites that Yeetline picked by hand and crawls politely, not the whole web. It answers from memory, on a page with no account, no cookie and no script.

What it keeps

Nothing about you. Not what you typed, not the address you typed it from, not that you were there, and no account, cookie or number that could tell two of your visits apart or link one search to another. Your words are read to answer them, and when they have been answered they are gone: they are written to no disk, no log and no database. What it does keep is the public pages it has crawled, and counts of how many searches it answered, by kind and never by what was asked. The reason for each thing it keeps, and how long, is on the search host's own /about page and in Yeetline's SECURITY.md.

Where your words go

Your words go straight to the search address (search. in front of this deployment's domain), an address that is not behind Cloudflare. On the search address there is no account, no cookie and no script. The page on the main address that holds a search box reaches you through Cloudflare; to leave it out, type the search address yourself.

That last sentence is the honest part. Cloudflare sits in front of the main address, so it can see that you opened that page, and it could change the page. It does not see the search: the box sends your words to the search address, which is not proxied, over a connection that Yeetline's front door passes through without opening, to a program that reads them in memory. The chat app and the database are not in that path. A search typed into the address bar of the main address, as a web address, goes through Cloudflare and reaches the app, which answers a plain "not here" page and keeps nothing; the visit itself is still seen.

Who may use it

The owner of this server chooses, in the settings only administrators can change:

• Everyone. The page has a search box and nothing else is asked of you. The program that reads your search is told nothing about who you are.
• Yeetline staff only. This is where it starts. You open Yeetfind from the chat app, which hands your browser a pass; the search address will not answer without one. The pass is the same for every member of staff for a few hours and says nothing about who holds it, so the engine cannot tell one member of staff's searches from another's. It does learn that whoever is asking is on the staff, and the staff are few enough that a time of day can point at a person. Staff mode is the less private of the two.
• Off. The search address answers that the engine is closed.

Where its pages come from

The pages it searches were read ahead of time by Spyder, the Stateless Polite Yeetfind Document Extractor and Retriever, on a schedule of its own and never because somebody searched. Spyder runs on the machine that runs the engine and, when the owner has set up more machines to help, on those. Every request it makes carries its name and a contact address, obeys the site's robots.txt and keeps the gaps between requests, whichever machine fetches. A helper is lent one address at a time, is never told a search and never holds the index; the program that answers searches neither changed nor gained a connection for it. What does change is who a site sees: the address of whichever machine fetched its page, and none of them is yours.

What it cannot promise

Said plainly, because a claim that is only partly true is worse than none:

• Whoever runs the machine can read a search while it is answered, from the program's memory, and could change the program to log. What stops that being quiet is the code, the published claims and the tests that fail when anything is logged, not a promise.
• Yeetline's app writes the route that sends traffic to the engine, and the machine's front door holds a certificate for every name under the domain, so a compromised app could send the search address somewhere else without a new certificate to show for it. Only a domain of the engine's own would take that away.
• On Yeetline's own host the domain's records are at Cloudflare. Whoever holds a domain's records can point a name somewhere else, and the checks the app makes from inside the machine cannot tell that from the real thing.
• In staff mode the pass travels through the chat app's connection, which may reach you through Cloudflare. Cloudflare could read the pass and cannot read a search.
• A site you click a result for sees your address, as with any link. Nothing here proxies your browsing.
• Anyone watching your connection can see that you connected to the search address, when, and roughly how long your words were.
• The crawler's address is the machine's, or, for a page a Spyder worker fetched, that worker's own machine's, so a site that is crawled sees one of them.

If a court or an authority asks

A search leaves nothing behind to hand over: there is no history, no account, no cookie and no address kept beside what was typed, and no log of it. The limit is the first one under What it cannot promise: a search is read in memory to be answered, and whoever controls that machine could read one while it is answered. A demand can reach only what exists, and nobody can know in advance whether a court could order a service to change how it works from now on. The account of what a demand can and cannot reach for the rest of Yeetline is on What this server can and cannot see.

What it is not

There is no account, no history, no suggestions, no trending list, no click tracking, no cached copy of a page and no "submit a URL". Each of those is a way to remember what someone searched, which is the thing this engine is for not doing.

Last changed 2026-10-19