1999.LOC

Signing a PDF normally means uploading it to a stranger or buying software. This runs in your tab; turn off your wifi once it loads and everything still works.

Sign it, fill it, cut text out of it, merge it, reorder it, split it. The toolbar across the top is what a click on the page does; the rail down the left is the document's own shape.

It rearranges. It does not re-render.

Pages are copied across whole: their fonts, their images, their form fields and their annotations come with them. Nothing is rasterised, nothing is re-compressed, and text stays selectable and searchable exactly as it was. A page that comes out of here is the page that went in, in a different place.

About deleting text, because this one matters.

Drawing a black rectangle over text is not redaction, and it is how governments and law firms have leaked documents for twenty years. The text stays in the file underneath the box: you can select it, search it, or hand the file to a chatbot and ask what the box is hiding. Free tools that offer “black out” almost always do exactly this.

So the delete tool here does not draw anything. Click a word and this finds the actual instruction in the file that drew it, and removes that instruction. The characters stop existing. Nothing is covered, nothing is flattened, and every other word on the page is still real selectable text, so you can delete one line out of a contract and the rest of it stays a normal PDF.

That only works when a word can be tied to one instruction. Some documents draw text in ways that cannot be, and rather than guess and cut the wrong sentence, the editor says so and points you at black out instead. That is the blunt fallback: the whole page is rendered to pixels with your boxes painted in and the original page thrown away, so the text really is gone, but the page stops being selectable text and the file gets bigger. Only the pages you black out are affected. If you black out a page, check the preview covers every character, because what your box misses is still readable as pixels.

The honest notes

• Nothing is uploaded. Static page, no endpoint, no account.
• This is not a digital signature. It draws your signature onto the page, exactly like signing a printout and scanning it. It is what almost everyone means by “sign this PDF”, and it is what most people are asked for, but it carries no cryptographic proof of who signed. If you need a certifying signature that Acrobat will validate, you need a certificate, and this is not that.
• Deleting is surgical; replacing is not re-flowing. Deleting removes the drawing instruction, so those characters leave the file. Replacing deletes the old and places new text at the same spot and size, which is what you want on a form or a name or a date. What it does not do is re-flow a paragraph: a PDF stores glyphs at positions, not sentences, so if your replacement is much longer than what it replaced it will run past where the old line ended instead of pushing the rest of the paragraph along. Nothing can honestly do that to an arbitrary PDF, and tools that claim to are usually rebuilding your layout and getting it wrong.
• Pictures go in at full resolution. A PNG or a JPEG is embedded exactly as it is, and nothing is re-compressed behind your back, so a photo straight off a phone makes a PDF the size of a photo straight off a phone. Anything else your browser can open (WebP, GIF, BMP) has to be repainted as a PNG first, which is lossless but throws away the original compression and usually makes it larger. The same picture placed twenty times is still stored once. HEIC from an iPhone will not open: browsers cannot decode it, and this page has nothing else to ask.
• Black out flattens the page. Only the pages you black out, and only when the surgical delete cannot safely be used, because covering text does not remove it.
• Make searchable reads a scan, and says it is reading. A scanned PDF is a photograph with no text in it, so “searchable” is optical recognition, not recovery: Tesseract reads the letters off each page and lays them back over the image invisibly, where you can select and search them. It downloads about eight megabytes the first time you ask, runs in this tab like everything else, and is slower than it looks: a page takes a few seconds. The result is every page as an image with a text layer, so a document that was already crisp vector text should just be saved normally, not run through this. And because every word came from a guess at a picture, proofread it before you trust it: recognition is good on a clean scan and poor on a bad one, and it only reads the Latin alphabet.
• It downloads about 700 KB the first time you open a file: pdf.js to render pages, pdf-lib to write them, plus font metrics for whatever your document uses. All served from this domain, because the security policy here forbids third-party scripts and a page handling your contracts is the wrong place to make an exception. The Pages tab shares it, so if you have used one the other is already cached.
• Pages are copied, not redrawn. Fonts, images, form fields, links and annotations travel with the page. This is the difference between a rearranged document and a photograph of one: the text is still text, still selectable, still searchable, and the file does not balloon.
• Turning a page is metadata, not pixels. A PDF page carries its own rotation, so a sideways scan is corrected by changing a number rather than by rotating an image. Nothing is re-encoded and nothing is lost. It also means the turn adds to whatever the page already had, which is what you want on a scan that was already sideways.
• What does not come across: the document's outline, its table of contents, its attachments, and any JavaScript in it. Bookmarks in particular point at pages that have moved, so carrying them over would produce a contents page that lies. They are dropped rather than mangled.
• Splitting into many files gives you one ZIP. A browser blocks a burst of downloads, so anything that produces more than one file comes back as an archive, which you can look inside with the archive tool.
• Large scans use memory. Everything is in the tab, so a 200-page scan may struggle on a phone.

Is this actually private?

Too big to email? Shrink it. Locked against copying? Unlock it. Just rearranging pages? The rail down the left side does that.

Hosted on Cloudflare · Made with spite for bloat
pdf to word | unlock a pdf | shrink a file | image tools | source

Nothing here is for sale, nothing is tracked, and there is no account to make. If something saved you an hour:
83TQcTwusSQ4WKbPQE5osrF3cR4GWe2zmcNWeozK6BSqHSaeLvjUVe476ouVwLKn1uVwEFcbJQvnme7W6dTV5SB93x45DEy
You can check that address here before sending anything.