JSON Minifier

Minification is the JSON equivalent of compilation: same semantics, fewer bytes, worse for humans. It matters in three situations — a payload embedded in a length-limited field, a document stored at scale where whitespace adds up, and any JSON crossing a network boundary.

What gets removed

JSON's grammar allows insignificant whitespace in exactly four places: before and after any of the structural characters { } [ ] : ,. That is the entire surface a minifier touches. Everything else — every character between the quotes of a string — is data and must survive byte-for-byte.

This tool parses the document and re-serialises it rather than running a whitespace-stripping regular expression. That distinction is important: a regex approach will corrupt a document containing "message": "a, b\n c", because the spaces and escaped newline inside that string are indistinguishable from formatting whitespace to a pattern that does not understand string boundaries.

The compression caveat

If your JSON travels over HTTP with gzip or Brotli enabled — and it almost certainly does — minification saves far less than the raw byte count suggests. Whitespace is extremely compressible: repeated indentation is exactly the kind of redundancy those algorithms excel at. In practice, minified-plus-gzipped and pretty-plus-gzipped are often within a few percent of each other.

Minification still matters when:

• the payload is stored uncompressed, in a database column, a key-value store or a config field;
• the transport cannot compress, such as a WebSocket frame, an MQTT message or a query parameter;
• there is a hard character limit, as with some serverless environment variables, SMS gateways or embedded systems;
• you are paying per byte at the edge and the volume is high enough for a few percent to be worth capturing.

A note on precision

Because minification goes through the JavaScript parser, the same big-integer caveat as the formatter applies: numeric literals above 2^53 − 1 will lose precision. If your document contains snowflake IDs or other very large integers, verify the output against the input before relying on it, or handle those fields as strings at the source.