URL Parser

Not as its main job. A dedicated URL Encode / Decode tool already exists on this site for standalone percent-encoding work on arbitrary text. This tool's own "decode query values" toggle only covers inline decoding of the query parameters it finds while parsing, as a convenience, so you do not need to copy values into the other tool just to read them.

No. Only values respect the decode toggle, as designed - keys are always shown exactly as they appear in the query string. This is a deliberate, narrow scope: percent-encoded keys are rare in practice, and always showing the literal key avoids silently merging two different keys that only differ after decoding.

Decoding is attempted with decodeURIComponent inside a try/catch for every value independently. If it throws, that one value falls back to its raw, still-encoded form rather than failing the whole parse - so one malformed parameter never breaks the rest of the output.

No - it uses the browser's own WHATWG URL parser rather than a hand-rolled one, since that implementation already ships correct in every browser and in Node. That means it inherits WHATWG behaviour, not strict RFC 3986: for example it will normalise a Unicode hostname to its punycode (xn--) form and collapse some path segments, which can look different from the exact characters you typed.