(Documentation in English)
What it does
Checks a JSON document on every keystroke and pins down exactly where it breaks. A badge switches between Valid JSON and Invalid JSON as you type. When the document is valid, a four-cell panel reports its size in bytes (UTF-8), its top-level type, its key or item count, and its maximum nesting depth. When it is invalid, the error names the line and column, and a second panel quotes the offending line from your input. Everything runs locally in your browser; nothing you paste is sent anywhere.
How to use it
- Paste or type JSON into JSON input. Tap Sample for a demo document, Clear to empty the box.
- Read the badge: Valid JSON (green) or Invalid JSON (red).
- If valid, read the stats panel:
- bytes - the document’s UTF-8 size, not its character count.
- type - one of array, object, string, number, boolean, null.
- keys (objects) or items (arrays) - the top-level count only.
- depth - maximum nesting; the top level counts as 1.
- If invalid, read
Line X, Column Y: <reason>, then the quoted line below it. - Copy input duplicates your text; Copy share link encodes it in the URL (
?j=...).
Examples
{"name":"CosmoDev","free":true,"tools":["json","base64"]}
Result: Valid JSON, 57 bytes, type object, 3 keys, depth 3. The strings inside tools sit two levels under the root, so the depth is 3.
A scalar is a complete JSON document:
| Input | Verdict | Stats |
|---|---|---|
123 |
Valid JSON | 3 bytes, type number, depth 1 |
true |
Valid JSON | 4 bytes, type boolean, depth 1 |
null |
Valid JSON | 4 bytes, type null, depth 1 |
An error pinned to its place:
{"a": 1}
extraResult: Line 2, Column 1 with the reason, and the panel below quotes extra.
How JSON validation works
A JSON parser reads the document once, left to right, and stops at the first character that cannot legally appear where it appears. The tool then maps that character’s offset to a position: it counts newlines before the offset for the line, and characters since the last newline for the column. Both are 1-based, so Line 2, Column 1 means the very start of the second line. The quoted panel prints that line from your input, so the excerpt always comes from your document, never from a generic template.
Four mistakes cause most invalid JSON in practice:
| Mistake | Example | Why it fails |
|---|---|---|
| Trailing comma | {"a":1,} |
after a comma the parser expects a key, and } is not one |
| Single quotes | {'a':1} |
strings and keys must use double quotes |
| Unquoted key | {a:1} |
keys are strings, and strings are quoted |
| Comment | {"a":1} //x |
JSON defines no comments at all |
None of these is a matter of style. The JSON grammar (RFC 8259) allows exactly one spelling for each, and the parser enforces it. That strictness is the point: a document that validates here parses identically in every conforming parser, in every language.
Good to know
- First error only: the parser stops at the first problem. Fix it, re-validate, and the next one surfaces.
- Engine differences: when the underlying engine reports no position, the verdict still shows the raw message.
- Shareable: your input is encoded in the URL (
?j=...), so a link reproduces the exact view. - Private: safe for configs, credentials, and API payloads.
- Related tools: JSON Formatter, JSON Minify, JSON ↔ CSV Converter.