Why diff JSON semantically?
A plain text diff of two JSON files drowns real changes in formatting noise: reordered keys, different indentation, trailing commas moved around. This tool normalizes both documents first — consistent indentation, recursively sorted keys — so the diff shows what actually changed: values, added fields, removed entries.
How it works
Both sides are parsed locally (an invalid document is reported with its side and position),
re-serialized in a canonical form, then compared line by line with the same algorithm used by
version control tools. The result is a unified diff — - for removals, + for additions,
with three lines of context — ready to paste into a ticket or a code review.