All the text tools, listed

The directory

What the text tools on this site are

483 routes, each doing one job to one piece of text. There is no tool on this page itself. It is the index, and its purpose is to answer two questions before you paste anything: which tool does what you need, and what happens to your text when you use it.

Most directories of text tools answer only the first. That works when every tool on a site behaves the same way. It does not work here, because these 483 split into groups that behave very differently, and the difference is not cosmetic.

The important split

Which text tools send your text, and which never do

83 of the 483 run entirely in your browser. Nothing is uploaded, there is no limit on length and no limit on how often you use them, because no request is ever made. Close the tab and there is nothing to have kept.

345 send your text to our server, because they need a language model to read it and there is no way to do that on your device. Those carry a limit of 5 runs per rolling 24 hours and accept between 1 and 12,000 characters per run.

You will find directories of text tools that promise your data never leaves your device, full stop. They can say that because every tool they host is client-side. We cannot say it, because it would be true of 83 tools here and false of 345, so the split is published instead.

Workflow

How to find the right text tool in three steps

The fastest route to the correct tool is usually to ask what kind of job it is before searching by name.

Step 1

Decide whether the job is mechanical or judgement. Stripping characters, converting case and reformatting JSON are mechanical. Rewriting, drafting and assessing are judgement.

Step 2

If it is mechanical, pick from the 83 local tools. They are free, unmetered, instant and they never see your text. There is no reason to spend an AI run on work a fixed rule can do.

Step 3

If it is judgement, pick the tool whose name matches the task most exactly, because the instruction sent to the model is built from the route name. A generic rewriter and a cover letter rewriter genuinely behave differently.

Runs on your device

The 83 tools that work in your browser

Twenty-four distinct operations across those routes, several with real options rather than a single fixed behaviour.

Cleaning

Hidden character removal, accent and diacritic stripping, space removal, whitespace minifying, line break handling, duplicate line removal, and a general cleaner with switches for normalizing spaces, collapsing blank lines and straightening quotes and dashes.

Converting

Case conversion with Title, UPPERCASE and lowercase, Unicode normalization across NFC, NFD, NFKC and NFKD, URL encoding and decoding, and px to rem conversion with a configurable root font size that defaults to 16.

Structure

JSON minifying and prettifying, conversion of delimited text to a JSON array including a custom delimiter option, HTML stripping, Markdown stripping and formatting, text splitting and find and replace.

Counting and checking

Word counting, token counting, and the watermark scanner that reports 14 invisible characters and 17 AI writing patterns.

Em dashes

Two separate routes, one that removes them and one that replaces them with a string you choose, defaulting to a plain hyphen.

Runs on our server

The 345 tools that use a language model

160 rewriters, 134 checkers and analyzers, and 51 generators. Rewriters take finished text and change it while keeping the meaning. Generators take a brief and produce something new. Analyzers read text and describe it without altering it.

All of them share one allowance of 5 runs per rolling 24 hours, so a rewrite and a detection check come out of the same five. The window rolls forward from when you used each run rather than resetting at midnight, so they return one at a time. Analyzers are also held to a rule worth knowing: they are instructed never to claim certainty about authorship or which model wrote a text, only to describe observable signals and their limits.

Files, not text

The image and video routes

38 routes handle watermarks in images and video rather than text. They work differently from everything else here, because a file has to be uploaded to our media workflow for the check or the edit to run at all.

That means the local-processing promise does not apply to them, and neither does the 12,000 character limit, which is a measure of text. If you arrived looking for a tool to take a logo out of a picture, those are the routes you want, and they are not among the 83.

Limits

What no limit means, and where it stops being true

On the 83 local tools it means what it says. No account, no daily cap, no character ceiling, and nothing counted, because there is nothing to count when no request leaves the page. Paste a hundred thousand characters and the only constraint is your browser.

On the 345 AI tools it is not true at all. Five runs per rolling 24 hours, and 1 to 12,000 characters per run, with anything longer refused before the model sees it. There is still no account and no payment involved. It is one honest limit rather than a limit that appears once you are invested.

Straight answer

Are these text tools private?

83 of them are, completely. Nothing is transmitted, so there is nothing to log, and that is a property of how they are built rather than a policy anyone has to keep.

345 of them are not, in the sense that your text is genuinely sent to our server so a model can read it. Nothing is written to storage there and the only record kept is a run counter holding a number and an expiry, but the text does travel and it would be dishonest to imply otherwise. If a document is sensitive, use a local tool, or do not paste it at all.