JSON, Base64 & Regex Utilities

JSON Formatter is for the one-line response curl hands you and the minified config someone committed. Paste it and it pretty-prints with 2- or 4-space indent, or minifies it back down before you save. It also normalizes whitespace, so reformatting before a commit keeps the diff clean instead of a wall of changed lines.

Base64 is for the middle segment of a JWT when you want to read the claims without a library, data URIs (data:image/png;base64,…), and the occasional blob someone pasted into a ticket. The encoder handles UTF-8 properly: native btoa() throws on anything outside Latin-1, which is what bites people the moment an accented character or emoji shows up.

Regex Tester is for validating input before it reaches a backend, or picking fields out of log lines. Matches and capture groups highlight as you type, and the error text tells you why a pattern is malformed instead of just saying “invalid”.

• JSON is stricter than JS. Single quotes, trailing commas, unquoted keys and comments are all invalid. Duplicate keys do not error: JSON.parse keeps only the last one.
• Numbers lose precision. JSON numbers are IEEE-754 doubles, so integers past 2^53 come back rounded when you re-serialize.
• Base64 has two dialects. URL-safe Base64 swaps + and / for - and _. Feeding one into the other produces garbage with no error.
• Regex can hang. Nested quantifiers like (a+)+ on a long non-match trigger catastrophic backtracking and freeze the tab.
• Lookbehind isn't everywhere. (?<=…) works in Chrome but throws in Safari before 16.4, so a pattern that tests fine on your machine may break someone else's.

Pull a response from staging with curl, drop the JSON in the formatter to see its shape, check a field's format against the schema in the regex tester, then Base64-encode a header for the next request. Three tabs of small work, none of it leaving the browser. The one thing to watch: all three are synchronous, so a multi-megabyte JSON payload or a pathological regex will stall the tab: split big inputs first.

All three utilities are thin wrappers around abilities the browser already ships. JSON parsing and serialization come from the native JSON object, Base64 conversion sits in the platform's btoa and atob functions, and regular expression matching belongs to the RegExp core described by the ECMAScript specification. Nothing here needs a server, because every operation follows a published web standard and executes inside the browser engine itself. Your data is read and written only in the tab's memory, and no network request carries it away. That is the point of a client-side tool: the machine doing the work is the machine you already trust, and the only cost is the CPU and RAM the tab uses while it runs.

The tradeoff is honest. Heavy work uses your CPU and RAM, and a multi-megabyte JSON document can stall the tab.

Each tool leans on a different specification. The JSON formatter validates and pretty-prints against RFC 8259, the document that defines the JSON data interchange syntax, so it knows exactly which constructs are legal. The Base64 encoder follows the alphabet laid out in RFC 4648, including the padding rules and the URL-safe variant that swaps + and / for - and _. The regex tester uses the pattern grammar the ECMAScript specification describes, which is the same engine your JavaScript runs under. The three standards sit at different layers: RFC 8259 governs a data interchange format, RFC 4648 governs the encoding of bytes, and ECMAScript governs a programming language.

So a single quote is a syntax error that the formatter reports, because RFC 8259 only allows double quotes.

Reach for the JSON formatter when an API response or config file arrives as one long line and you need to see its shape, or before committing so that whitespace changes do not flood the diff. Use the Base64 tool when you want to read the middle segment of a JWT, inspect an embedded data URI, or decode a blob someone pasted into a ticket. Turn to the regex tester when you need to validate input before it reaches a backend or pull fields out of log lines with instant feedback on groups and errors. If a payload survives a round trip through both sides of the encoder without changing, the tool did its job.

The three overlap less than they seem. Choosing mostly comes down to what you are looking at.