The dev tools you settle for, done right.
Formatters, validators, converters, and playgrounds. Every one runs in your browser, respects your data, and doesn't look like it was built in 2008.
Format, diff, and query without your data leaving the tab.
Developers format JSON a dozen times a day, test cron expressions a dozen times a month, and throw CSVs into a converter whenever a PM hands them a spreadsheet. These are small daily tasks, the kind you do without thinking. The tools for them should be excellent. Most are garbage.
Most of the free ones are built on a model nobody's looked at since 2008. Server-side processing. Ad-stuffed chrome. Dark patterns aimed at your email address. Paste a JSON blob into jsonlint.com and it travels, unencrypted, to whatever server is hosting that domain this month. Format SQL in a web tool and the query gets logged somewhere you'll never audit. That query might reference production tables, column names that reveal your data model, or literal values with PII.
The technology to fix this has existed for years. WebAssembly runs real compiled engines right in your browser. The actual jq binary. A real SQL parser. The Markdown pipeline you'd otherwise run on a server. Plain JavaScript is fast enough to format gigabyte payloads on a mid-range laptop. There's no technical reason a formatter needs to round-trip your data anywhere. Server-side was just easier to ship in 2012, and nobody bothered to update the design. We did.
Where the old model breaks
- 01
Your data leaves the browser
The default model ships your paste to someone else's server before anything renders on screen. For a webhook payload or a prod config, that's backwards. The formatting happens after the exposure, not instead of it.
- 02
No dark mode, no keyboard shortcuts
A workflow built for someone who's never opened devtools at night, debugged from a phone on a train, or touched a keyboard shortcut. Nobody you've actually worked with.
- 03
Broken in quiet, specific ways
CJK characters mangled in tables. Timestamps silently clipped to seconds. A minifier that strips comments. A diff that ignores whitespace even when whitespace is what changed.
- 04
No output you can trust
Formatters that reorder your keys. Converters that guess wrong at types. Validators that say 'invalid' and leave you to hunt the broken field yourself. The kind of tool you stop trusting and keep using anyway, because the alternatives are worse.
The tax is bigger than it looks. Format JSON fifteen times a day and you've spent an hour this month waiting on ad overlays and server round-trips that don't need to exist.
Pick the tool by the task
Each of these replaces one specific bad experience: a JSON formatter buried under ad banners, a diff viewer that treats structured data like flat text. All eleven run the same way, in your tab, on your machine, and none of them ask for an account. The three you'll probably open first are Jqbin, tidysql, and Diffpad, because payloads, queries, and configs are where the round-trip model costs the most. Find the row that matches the job.
A few of the others look narrow until the day you need them. Oktal earns its slot the first time a code review turns on what a 644 permission mask actually grants, and Epochr the first time two log files disagree about whether a timestamp is counted in seconds or milliseconds.
These are small tools. They need to load fast, work correctly, and stay out of the way. Everything else on this page follows from one architectural decision: the engine runs inside your tab, so the data never gets a chance to go anywhere else. A formatter that can't phone home doesn't have to promise it won't. The architecture is the privacy policy.
Developer Tools
The jq playground developers actually keep open.
jq is the best way to slice JSON at the command line, but iterating on a filter inside a terminal is slow and disposable. Jqbin runs the real jq engine in your browser via WebAssembly, keeps a local history of every query you've written, and gives you a permalink you can drop into a coworker's DM. Twenty-plus curated recipes ship with it. Extraction, grouping, array work, the patterns you'll actually reach for.
Try it 02Format SQL in seconds, right in the browser.
Every online SQL formatter fails somewhere. Some strip comments without asking. Others mangle dialect-specific syntax. Most force one opinionated style and won't let you bend it. tidysql handles fourteen dialects. Postgres, MySQL, SQL Server, BigQuery, Snowflake, and the rest. Real control over keyword case and indentation, and a side-by-side diff that shows exactly what changed. That matters when you're cleaning a migration that has to go through review before it ships.
Try it 03The fastest, cleanest JSON formatter online.
Paste JSON, get it formatted before your finger leaves the V key. Syntax highlighting separates keys, strings, numbers, booleans, and null. Errors come back with exact line and column numbers. Minify mode compresses for transport, dark mode respects your system, and everything runs client-side. That matters when the payload you're inspecting is a webhook, a prod config, or an LLM response with user data you'd rather not hand to a third party.
Try it 04Paste. Compare. Copy the diff.
Diff tools built for flat text treat JSON and YAML like two walls of characters. Paste two blobs into diffpad and you get a structural diff that actually understands keys, arrays, and nested objects, with color-coded additions, deletions, and changes. The plain-text summary copies with one click. That's what you need when you're explaining a config change inside a pull request comment or a Slack thread.
Try it 05See exactly which cron field is wrong, and why.
Cron expressions fail quietly. A typo in the month field or a bad day-of-week spec won't throw an error. The job just doesn't run, and you find out when something you expected to happen didn't. lintcron validates each field on its own, tells you exactly which one is broken and why, translates the expression into a schedule you can read, and previews the next few runs so you can sanity-check the cadence before you commit it.
Try it 06Paste a messy table. Get a perfect one. Instantly.
Markdown tables look simple until you paste one with ragged columns or wide Unicode characters. pipefmt aligns pipe-delimited tables correctly and accounts for CJK characters that take up two visual columns. That's an edge case most formatters quietly get wrong. Live preview updates as you type, the output copies in one click, and the result pastes cleanly into GitHub issues, readmes, and anywhere else a table has to land legibly.
Try it 07CSV to JSON in one paste.
CSV-to-JSON is the conversion you run when seeding a database, scaffolding a prototype, or converting a spreadsheet a PM just handed over. Rowson auto-detects the delimiter (comma, tab, pipe, or semicolon), handles quoted fields and embedded newlines, and gives you four output shapes: arrays of objects, objects keyed by column, flat arrays, or pivoted. Pretty or compact. Copy or download. Nothing uploaded.
Try it 08Hex math that shows its work in every base.
Bitwise ops are still how flags, permissions, and low-level protocols work, and the math is miserable to do by hand. Oktal evaluates hex expressions with arithmetic and bitwise operators (AND, OR, XOR, NOT, left and right shifts), then shows the result in hex, decimal, octal, and binary at the same time. Good for debugging packed fields, decoding file permission masks, reading protocol specs, or just sanity-checking your mental math before a PR goes in.
Try it 09Build an RBAC matrix. Export it anywhere.
RBAC matrices usually live in a spreadsheet because nobody builds a dedicated tool for them, and from there they get hand-transcribed into Markdown for a README or wiki. It's a workflow nobody enjoys. Rolemat gives you a real checkbox grid (up to fifty roles by a hundred permissions) and exports to CSV, Markdown, or JSON in a single click. The Markdown is paste-ready for GitHub, which is where RBAC docs usually end up anyway.
Try it 10The Unix timestamp converter that tells you what you're looking at.
Unix timestamps come in four flavors: seconds, milliseconds, microseconds, nanoseconds. And log files almost never label which one they're using. Epochr guesses the unit from magnitude, converts to ISO 8601, local time, UTC, and relative formats, and copies anything in one click. The tool you leave open in a background tab when you're correlating events across systems that disagree, quietly, about precision.
Try it 11Token count, cost, and context window for GPT, Claude, and Gemini.
LLMs charge per token. Context windows are measured in tokens. And yet most developers are working blind about how many tokens their prompts actually burn. Toklen counts tokens for GPT, Claude, and Gemini, estimates cost per request, and shows context window usage as a live progress bar. Everything runs in your browser, which matters when the prompt you're measuring is internal IP you wouldn't want to paste into a third-party endpoint.
Try itExplore more categories
See all 43 tools.