What is DuckFind?
DuckFind is a search engine and article reader for old computers. The modern web is megabytes of scripts, styles and fonts behind TLS connections a vintage browser cannot even open -- so DuckFind fetches today's pages on the server and hands your machine clean HTML 3.2: no scripts, no stylesheets, images converted to small GIFs. It renders on anything from a System 7 Mac or a Windows 3.1 box to an Apple II with a text browser.
Search is powered by DuckDuckGo. Also included:
- Wayback reader
- news portal and personal feed reader
- street maps with directions
- weather forecasts
- calculator and unit/currency converter
- dictionary
- PDF viewer
- file-download proxy
- Hacker News with comment threads
- translator
- gopher
- anonymous FTP browser/downloader
- Wikipedia lookup
- Project Gutenberg ebook reader
- Wiby classic-web search
- YouTube video playback
- YouTube search
- AI answers (!ai)
- bang shortcuts (try !help)
Inspired by FrogFind (frogfind.com) -- an independent, open-source implementation you can run yourself.
Privacy
Searches and pages are fetched by the server on your behalf, so websites and search engines see DuckFind's address, not yours. Fetched pages and images live briefly in a server cache keyed by URL -- never by visitor -- and are deleted within a month at most. Rate limiting stores a salted hash of your address, never the address itself.
The one exception: My Feeds saves your feed list under a salted hash of your reader code, and it persists as long as you revisit it at least once every ~400 days -- everything else above is deleted within a month.
The optional !ai shortcut sends that question -- and nothing else -- to Anthropic to generate the answer.
The optional !watch shortcut is handled by a separate server (reached over Cloudflare) that fetches and converts video for old machines -- unlike the rest of DuckFind, that path is not guaranteed log-free: it keeps no access logs of its own, but Cloudflare sits in between and may log traffic through it. (!ytsearch is unrelated -- it runs a metadata-only search locally on this server, no relay hop.)
This site keeps no logs of what you search or read -- no web-server access logs, and no logging proxy or CDN in front.
One inherent limit: vintage browsers speak plain HTTP, so the networks between you and this server can observe that traffic in transit -- the price of working on old machines. A modern browser can use the encrypted https:// address instead.
Limits
Limits keep DuckFind available for everyone (and keep its search backend happy). Per visitor:
- 40 searches per minute
- 90 article reads per minute
- 200 images per minute
- 60 news pages per minute
- 30 map requests per minute
- 15 PDF requests per minute
- 20 downloads per minute
- 20 FTP listings or downloads per minute
- 10 AI answers per hour
- 40 video conversions per 5 minutes
- 20 video searches per minute
- 80 video status checks per 5 minutes
Shared by all visitors: a daily budget of 5000 new searches and 500 AI answers. When it runs out, recently searched terms keep working from cache and everything resets at midnight UTC.
Fetches are capped in size and time, so very large pages and images arrive trimmed rather than not at all. The download proxy accepts files up to 50 MB.
Meet Mallard
Mallard is DuckFind's detective duck -- he does the legwork of fetching, stripping, and converting the modern web so your old browser never has to deal with it directly.
news · my feeds · maps · reader · guestbook · shortcuts
DuckFind -- the modern web in plain HTML, for vintage browsers
settings · about · search powered by DuckDuckGo
no ads · no tracking · no logging