JS Object ↔ JSON

Accepts the loose syntax you write in actual JS: unquoted identifier keys, single quotes, trailing commas, line comments. Emits canonical JSON (or, in the reverse direction, a clean JS literal that matches what you'd type by hand). Useful for turning copied dev-console output into something a JSON parser will accept, or for prettifying API responses for inclusion in source code.

Parsing goes through JSON5 rather than new Function, which is the difference between a converter and an arbitrary code execution hole. Handing pasted text to the JavaScript engine would run whatever was in it, and object literals copied from a console are exactly the kind of text people paste without reading. The trade-off is that only data survives: a function, a new Date(...), a regular expression literal or a computed key is a parse error here rather than being evaluated into something else.

JSON5 also accepts values JSON has no way to express. NaN, Infinity, hexadecimal numbers like 0xFF and leading-decimal forms like .5 all parse, but JSON has no syntax for the first two, so they come back as null and the round trip is not always lossless. Numbers are another quiet edge: values beyond the range of a double lose precision on the way through, which is why a 19-digit identifier arrives back subtly changed. Keep long IDs as strings.

In the JS direction, keys that are valid identifiers are left unquoted and everything else is quoted in the style you pick. That output is a literal, not a statement, so assigning it needs the usual wrapper, and pasting a bare object at the start of a line makes the parser read the braces as a block.

Comments are dropped. They are not part of the data model on either side, so a heavily annotated config comes back clean but bare.