A straightforward approach to privacy.

Files and text stay in your browser.

The image and PDF tools read the files you choose on your device and hand you the results through your browser. The JSON formatter processes pasted text in the current tab. KeepSimp’s application does not upload this content or store it on a server, and it saves none of it in browser storage except as described below. Otherwise, leaving or refreshing a tool page clears what you were working on. Downloads are saved wherever your browser normally saves them.

What is kept only when you ask.

The PDF editor can keep four things in this browser’s own storage, on this device. Each is kept only after you turn it on or type it, and none is sent anywhere. The name you sign your comments with is kept once you type one. A signature or your initials are kept when you switch on “Remember this signature on this device”, until you press “Forget it”. A draft of a PDF you are editing (the file, the pictures you placed, another PDF whose pages you added, a font of your own if you used one, and your changes) is kept when you switch on “Keep a draft on this device”, until you close the file, throw the draft away or switch it off. The certificates you add as ones you trust, to check signatures against, are kept when you switch on “Remember them on this device”, until you switch it off. Do not turn these on at a computer other people use. A certificate you sign a PDF with, and its password, are read on this device for that download only: they are never stored or sent. You can also remove them through your browser’s site-data settings.

The voice recorder keeps a take in this browser’s own storage, on this device, only while “Keep a draft while recording” is on (it starts off every time), so that a closed tab does not lose it. The draft is deleted when the take is finished or you delete it, and it is never sent anywhere. The microphone is used only after you press Record and the browser asks you, and only until you press Stop.

The Markdown editor reads the files you open on this device. It keeps nothing of them in the browser's storage unless you switch on “Keep a draft on this device”: then your open documents, the pictures you added to them and their versions are kept in this browser's own storage (IndexedDB), on this device, until you switch it off, which deletes them. Where your browser lets a page write a file or a folder you chose (Chrome and Edge), Save writes back to that file and nowhere else, and only while the page is open. Pictures that a document links from other sites are fetched by your browser to show them in the preview, as any page fetches its pictures; the preview options can leave them out. Links are never checked over the internet.

The video editor writes each video it makes to this browser’s own private storage on this device while it makes it, so that long videos do not fill the memory. That copy is deleted when you remove the video or its result is replaced, and when you close the page; if the browser stops before it can, the next time the video editor opens. It is never sent anywhere. In a private window, where that storage may not be offered, results are held in memory instead.

The website still loads from a host.

Page HTML, scripts, fonts and public images load from the website host, Vercel. A host receives ordinary request information such as an IP address, the requested URL and browser headers. Browser-local processing does not mean website visits are invisible to the hosting infrastructure. This application does not include advertising scripts.

Website traffic, without analytics cookies.

On the production site, Vercel Web Analytics counts visits and page views without analytics cookies. It sends page URLs, referrers and basic browser, device and approximate location information to Vercel for aggregate traffic statistics. We do not send custom events, file contents, pasted text, passwords or memory readings to analytics. See Vercel’s Web Analytics privacy details.

A couple of display preferences.

The app stores your chosen light or dark theme in local storage and may store the desktop sidebar state in a cookie. Those preferences contain no selected files or pasted text. You can remove them through your browser’s site-data settings.

When an AI agent uses the tools.

The KeepSimp bridge, which lets agents such as Claude Code or Codex use these tools, runs on your computer. It opens KeepSimp in a browser of its own, hidden, and refuses every other address. It reads the files your agent names from your disk, only inside the folders the agent works in, hands them to that browser and saves the results beside them. Neither the bridge nor the page sends your files or their contents anywhere. What your agent does with an answer is up to it and the service it runs on. An agent in your browser that uses the tools on the agent page sees only the files you add there.

A copy of the tools for offline use.

Once you open the tools, your browser keeps a copy of KeepSimp’s own pages, scripts, fonts and codecs, about 37 MB, so they work without a connection. That copy contains none of your files or text. It is replaced when a new version is published, and you can remove it through your browser’s site-data settings. What recognises the words of scanned pages (about 1 MB, and a few megabytes for each language) is not part of it until you first recognise text in that language; it is then kept the same way. It too holds none of your files, and the words it finds go only into the PDF you are editing. What takes background noise out of sound (about 3 MB), and Whisper, which transcribes speech (about 107 MB, or 280 MB for the larger model), are kept the same way the first time you use them; the sound, its noise and its words are never sent anywhere.

Keep the limits in mind.

Browser extensions, clipboard synchronization and device software can have their own access to content. Use a browser and device you trust. Keep original files: image exports can remove metadata, profiles, transparency or animation depending on the chosen format. Follow any requirements that apply to the information you handle.

Future tools follow the same direction.

Video and audio tools remain marked as coming soon. Their data handling will be documented when they are implemented. If a future tool needs a different approach, the interface and this page must explain it before a user provides content.