Grouped by the situation, not the format
This category is not a theme so much as a drawer. The tools in it are the small conversions you need while something else is the actual task, so they are listed here by the situation you are in when you reach for them.
“There’s an opaque string in this log and I don’t know what it is.”
If it looks like 550e8400-e29b-41d4-a716-446655440000, the
UUID decoder tells you the version and, for v1,
v6 and v7, the timestamp it was created at. If it looks like SGVsbG8= or ends
in ==, it is Base64 —
decode it. If it is a bare ten-digit number
starting with 17, it is a
Unix timestamp from roughly now.
“I need an ID.” The UUID generator does v1 and v3 through v7 and will produce thousands at once for seeding a test database.
“This timestamp is wrong by a factor of a thousand.” Seconds or milliseconds — the timestamp converter handles both and shows you which one you gave it.
“I need this data URI as a real file.” Or the reverse: Base64 to image.
“What colour is that exactly?” The colour picker for converting between notations, the image colour picker for reading a value out of a screenshot or a mockup.
“This needs to be snake_case and it’s camelCase.” Case converter.
“What is 0xDEADBEEF in decimal, and what happens if it’s signed?” Number system converter, which unlike most handles negatives and two’s complement.
Identifiers leak more than people expect
The one cross-cutting thing worth saying here, because no single tool page owns it: several of these formats carry information you did not intend to publish.
UUID v1 embeds a timestamp and the machine’s MAC address. That was the original design — uniqueness guaranteed by “this machine, at this moment”. It also means a v1 UUID in a public URL tells anyone who cares exactly when the record was created and gives them a stable identifier for the server that made it. You can confirm this yourself by pasting one into the decoder.
UUID v7 embeds a Unix timestamp too, deliberately, because sorting by ID and sorting by creation time being the same thing is enormously useful for database index locality. The cost is the same disclosure: v7 IDs are ordered, so exposing them publicly reveals creation times and lets someone estimate how many records you create per hour by watching two of them.
v4 is the one that reveals nothing — 122 bits of randomness and no structure. That is why it remains the default for anything user-facing, and why “use v7 for the primary key, v4 for the public identifier” is a reasonable rule for the same table.
Base64 reveals everything, instantly, to anyone. It is not obfuscation. A Kubernetes secret is Base64-encoded so that binary values survive YAML, not to protect them.
Everything on this page runs locally, which for this category matters more than most: production identifiers, tokens and log fragments are exactly the sort of thing that should not be pasted into somebody else’s server to be decoded.