A jar of small tools, and no server.

Most of the utilities a search turns up will happily take whatever you paste and send it off to a server. That's a bad deal when the thing on your clipboard is a production response with someone's personal details in it. None of this work needs a server, so ToolJar doesn't have one. Open your network tab and you won't find what you pasted in any request, or go offline entirely and keep using it. The privacy page spells out the one exception and how to check the claim yourself.

28 tools on 6 shelves. The list below is read straight out of the same registry the menus and the sitemap use, so it always matches what is on the site today:

JSON tools: Format, validate, compare and convert JSON.

JSON Formatter · JSON Validator · JSON Tree Viewer · JSON Diff · JSON Path Finder · JSON to CSV · CSV to JSON · PHP Unserialize · PHP Serialize

JWT tools: Decode, verify and generate JSON Web Tokens.

JWT Decoder · JWT Verifier · JWT Expiry Checker · JWT Generator · Base64URL Encoder

Time & date tools: Convert timestamps, read cron, compare zones.

Epoch Converter · Cron Expression Explainer · Time Zone Converter · Duration Calculator

Keys & IDs: Generate secure tokens, passwords and UUIDs.

Password Generator · UUID Generator

Code formatters: Format SQL, HTML, CSS, JavaScript and PHP code.

SQL Formatter · HTML Formatter · CSS Formatter · JavaScript Formatter · PHP Formatter · Markdown to HTML

AI tools: Count tokens and hide secrets before you prompt.

Token Counter · Prompt Secret Scrubber

More go in as I need them. The rule for anything that gets added is the same: it has to work without a server, and it has to do the job better than the first search result does. Usually that means being specific where the alternatives are vague: naming the line and column that broke, naming the cron field that is out of range, or saying outright that the hour you typed does not exist because the clocks moved.

A static React and TypeScript app, kept small. It is an independent project, not a company, so there is nothing here to upsell you and no business model that needs your data. There is no backend and no database either; every page is rendered to plain HTML at build time, so a tool is on screen before much JavaScript has had to run.

The parts that had to be exactly right are written from scratch, with no dependencies. The JSON engine (parser, formatter, diff and statistics) is hand-written rather than wrapped around JSON.parse, mostly so its error messages can name the exact line and column, which the browser's own parser doesn't do consistently. So are the CSV reader, the JWT signing and verification (WebCrypto, secret never leaving the tab), the cron parser and the time zone arithmetic, which leans on the browser's own Intl data rather than a hardcoded table of offsets that would be wrong within a year. The code formatters are the exception: they wrap the established libraries for each language, because a hand-rolled JavaScript minifier can easily change what someone's code does.