Online JSON Diff Tool

Paste one document into each box, or drop the files onto them. The comparison runs as you type: the diff tree marks every added, removed, edited and retyped node in its own colour, each with its key path. Nothing leaves your browser — the documents are never uploaded.

A text comparison looks at lines. Reformatting a document, changing its indentation or writing its keys in another order produces hundreds of line differences even though no value changed. A structural comparison parses first: an object is a set of keys, an array is a sequence of items. What you get back is only the values that actually differ.

No. In the JSON standard the keys of an object are unordered, so {"a":1,"b":2} and {"b":2,"a":1} are the same object. The tool reports no difference between them, but it does tell you when the only thing that changed was the order — useful confirmation that a rewritten file is otherwise untouched.

By default yes, but insertions and deletions are detected, so adding one record at the top does not make everything below it look changed. For lists of records switch to key matching: items are paired on an identity field such as id, uuid or key. A record pulled to the top then reads as moved rather than as a deletion plus an addition. You can also type the field name yourself.

Because in JSON one is a number and the other is a string, and to most code consuming the document that is a real difference: a field that returned a number in one version of an API and returns text in the next is exactly what a comparison should surface. If the distinction does not matter to you, turn on the number-and-text option — the tool then also tells you that the patch no longer reproduces the target document exactly.

No, and the tool keeps them apart. A key whose value is null exists and holds an empty value; a key that is absent is not defined at all. The generated JSON Patch preserves the distinction too: one produces a replace, the other a remove or an add.

A JSON Patch is a list of operations that turns one document into another, and it is itself JSON. You can send it in the body of an HTTP request, write it to an audit log, or keep it in version history as the change between two configuration revisions. The tool writes the patch, then applies it to the first document and checks whether the result is the second one, showing the verdict on screen.

The standard does not forbid it; most readers keep the last value and some keep the first, so the same document can quietly be read two different ways. The tool accepts such a document but does not swallow the repetition: it warns you, naming the key and the line it repeats on.

The comparison stops and the reason appears under that box with a line and column number: an unclosed bracket, a trailing comma, single quotes, an unquoted key or a malformed number are each named separately. Instead of the browser's own wording you get a sentence saying what was expected and where.

No. Parsing, comparing and patch generation all run in your browser; the documents are neither uploaded nor stored. That makes the tool safe for production configuration, API responses holding customer data and personal records.