Every tool, in one place

Local processing is a statement about the code rather than a slogan. When a tool page says it runs in your browser, the work is being done by APIs the browser already ships: Canvas and the image decoder for pictures, pdf-lib and PDF.js for documents, Web Crypto for hashes and keys, and FileReader to pull your bytes into memory in the first place. There is no upload endpoint in this codebase for those tools to send a file to, which is a stronger guarantee than a promise not to look at what arrives.

The consequences are practical rather than abstract. A conversion keeps working after your connection drops, because nothing was travelling over it. Speed is bounded by your own processor instead of by a queue or a rate limit. And there is nothing to delete afterwards — no retention window to wait out, no request to file, no account to close — because the file was never written anywhere except that tab’s own memory, which the browser releases when you close it.

What the site does keep stays on your device: at most three values in localStorage, holding your light or dark theme choice, the list of tools you opened most recently so the homepage can offer them again, and your answer to the cookie banner. None is transmitted to us and none is readable by anyone but the device that wrote it. There is no account and no advertising cookie set by this domain. Google Analytics is the only third-party script, it is off until you accept, and advertising storage stays denied even after you do.

249 of the 250 tools work this way. The remaining 1 is Instant Indexing, and the reason is worth stating plainly rather than burying: it asks search engines to crawl URLs, and the IndexNow endpoints it talks to return no Access-Control-Allow-Origin header. A browser is permitted to fire that request but forbidden from reading the reply, so without a relay the tool could not tell you whether your submission had worked or silently failed. Our server makes the call and returns only the status code; what you enter is forwarded to complete that one request and is not logged, stored or reused. Its Google tab needs no relay at all — your service-account key is signed inside the browser and sent only to Google.

The search box matches tool names, their descriptions and a set of keywords kept alongside each record, so typing jpeg surfaces the JPG tools and typing compress surfaces both the image and the PDF compressors. Exact name matches rank above partial ones, and an empty box simply shows the whole library.

Tools are named after the job rather than the technique — WebP to PNG, Merge PDF, Remove Duplicate Lines — so the directory reads as a list of tasks you might have. If you know the category but not the tool, the six hubs above each carry their own indexable page with everything inside them.

Once you have found one tool, its own page links to six related ones and to the rest of its category, which is usually the quickest route to the neighbour you actually wanted: the resizer beside the compressor, the word counter beside the case converter, the decimal-to-fraction tool beside the fraction calculator.

If nothing fits, the contact page reaches a person rather than a ticketing queue. New tools arrive in batches, and a request that describes your input format, the output you need and what you are trying to accomplish is far more likely to get built than one that names a feature.