When JSON breaks, the error message is often blamed for being vague. The bigger problem is usually that the data is being inspected in the wrong form. A dense one-line payload is hard for a human to reason about. Formatting first turns structure into something your eyes can follow.
Step 1: preserve the raw payload
Копировать the original response or file before editing it. Debugging becomes much harder when you “fix” several characters and then cannot remember what the service actually returned.
Step 2: parse, do not just indent
A real formatter should parse the JSON. If parsing succeeds, pretty-printing can safely add indentation. If parsing fails, the useful result is the error—not a cosmetically indented invalid string. Common failures include trailing commas, single-quoted strings, unquoted property names and missing closing brackets.
Step 3: read around the error, not only at it
Parsers often report where they finally became unable to continue, which can be slightly after the actual mistake. If the error points at a closing brace, look at the property immediately before it. A missing comma on the previous line is a classic example.
Step 4: check the data shape
Valid JSON can still be wrong for your application. Confirm whether the top level should be an object or an array, whether field names match the expected schema and whether numbers have accidentally become strings. Syntax validity is only the first layer of correctness.
Example: the harmless-looking trailing comma
{
"name": "Ada",
"active": true,
}Many programming-language object literals allow a trailing comma. Strict JSON does not. Removing the comma after true makes this particular example valid.
Minify only after validation
Minification is useful for transport and storage, but it is a poor debugging format. Keep the readable version while investigating, then minify the validated result if your workflow benefits from smaller text.
Be careful with sensitive API responses
JSON often contains tokens, email addresses, internal IDs or customer data. Use local/browser processing for sensitive material when available, redact secrets before sharing debugging examples, and never paste production credentials into public issue trackers.
A good stopping point
You are done when the JSON parses, matches the expected shape and still represents the same information as the original payload. “The formatter turned it green” is not enough if the application expects a different schema.