In-Tab Tools

Unicode shapes that survive the paste. They are characters, not a font, so they travel with the text into any app that accepts text.

• Cursive font generator→Type once and every cursive variant appears below. Take the one that suits the bio you are writing.
• Gothic font generator→Fraktur and blackletter characters, ready to go straight into a profile.
• Discord text formatting→Write the message and copy the markdown. No counting asterisks to work out which one you missed.
• Discord timestamp generator→Pick a moment, read how each format will look to someone else, copy the code.
• Ambigram generator→Two words, one drawing: it reads one way up, and the other word when you turn it around. Rotate the preview and read it yourself — every letter pair is scored, and the bad ones are named.
• Small text generator→Three different kinds of small text. They behave differently, and one of them is missing most of its letters.
• Glitch text generator→Stacked combining marks, with the intensity under your control. This is the style most likely to break something.
• Bubble text generator→Type text, get circled bubble letters back — outline covers the full alphabet and digits, filled does not.
• Subscript generator→Type text and get real subscript characters back, not a smaller font, with the letters Unicode never gave a subscript form to.
• ChatGPT watermark remover→Paste text from ChatGPT and it names every hidden character in there, then hands back a clean copy.
• Remove invisible characters→Paste text and it names what is hidden in there, by category and count, then hands back a clean copy.
• Invisible character→Eleven blank characters, each one named, with the one thing that decides whether it works: is it whitespace or not.
• Invisible character for Discord→The one that works in a Discord name is U+3164. Here is why the one you tried did not, and what to do when a blank nickname resets itself.
• Invisible text→Blank text that pastes as real characters — for an empty message, a gap in a bio, or indentation that does not collapse.

You give it a number of kilobytes and it compresses until the file fits. The limit you are fighting is a byte count, so that is what the tool takes.

• Compress image to 20KB→Drop a photo in and it comes out at 20KB or under. This is the tightest target here, so expect visible quality loss.
• Image compressor to 20KB→Built for the case where you have thirty files and every one of them has to come in under 20KB.
• Compress image to 50KB→Pick the size, and the quality is worked out for you. The file comes out at 50KB or under.
• Compress image to 100KB→Enough headroom that most photographs reach the target on quality alone, at their original resolution.
• Compress image to 200KB→The most forgiving target on the site. Most photographs get there without any visible cost.

Both directions, parsed on your machine. No model reads the document and no page count is billed.

• PDF to MD→Drop a PDF in and Markdown comes out. Nothing is uploaded and nothing is billed per page.
• Markdown to PDF→Renders your Markdown, then hands it to the browser's own print engine. The PDF is written on your machine.
• PDF metadata remover→It shows you what the document is carrying first. Deleting is a second, deliberate press.
• Remove metadata from PDF→Author, producer, creation and modification dates, XMP, and who left which comment. Listed first, deleted on request.
• Remove PDF metadata→The version you send should not say who wrote it, on what, or when you last touched it at 2am.
• Compress PDF to 200KB→Drop a PDF in and it is re-encoded down toward 200KB, image by image. Nothing leaves your browser.
• Compress PDF to 100KB→A common upload cap on application forms. This re-encodes the images inside your PDF until it fits, or gets as close as the content allows.
• Compress PDF to 500KB→The most forgiving of the three targets: there is enough room left that a scanned document usually still reads clearly afterward.

It lists what an image is carrying before it deletes anything, and the pixels are copied across untouched. Reading is the default; deleting is the deliberate act.

• Metadata remover→It tells you what is in the file first, then takes it out. Nothing is uploaded.
• Remove metadata from image→Built for the case where you are about to post a batch of images and would rather they said nothing about you.
• AI metadata remover→Image generators label their output. This removes the labels that live in the file, and tells you plainly which ones it cannot reach.
• Claude watermark remover→Claude signs the images it generates with a C2PA provenance record. This finds it, shows you it, and rebuilds the file without it.
• C2PA remover→Content Credentials are a block of data on the file. Removing it is a byte edit, and this shows you the block before it goes.
• C2PA checker→Drop an image in and see the provenance manifest it is carrying, if it has one, along with what that manifest says about where the image came from.
• SynthID remover→Google puts two marks on a generated image. This takes off the one that is a file record, byte for byte, and leaves your pixels alone. The other one lives in the pixels and does not come off here.

Overlays an evenly spaced grid on a reference image, so you can copy or scale it one square at a time instead of judging the whole picture at once.

• Grid maker→Drop a reference photo in and it comes back with a drawing grid over it, ready to copy square by square.

Convert subtitle formats in your tab. Drop an SRT file and choose plain text or web format, nothing is uploaded.

• SRT to TXT→Drop a .srt file in and read the plain text it was hiding — the numbers and timecodes are gone, the words are not.
• SRT to VTT→The two formats are almost identical. This adds the one line and fixes the one punctuation mark that keeps a browser's video player from reading an .srt file directly.

Process audio files in your browser. Reverse a track or work with its samples without uploading anything.

• Audio reverser→Drop in a sound file and hear it played backwards straight away, then download the reversed WAV. Decoding happens in this tab, nothing is uploaded.