Base64 decode
Base64 decode. Runs on your device, nothing is uploaded.
Base64 decode strips whitespace, checks every character against the standard alphabet, and rejects a length whose final group cannot be a valid base64 tail after padding. An unrecognised character stops the run, with its exact spot in the cleaned text given. Text decoding to valid UTF-8 shows on screen; an image or compressed blob comes back as a decoded.bin download instead.
Questions
Is padding required?
No, trailing = padding is optional; the tool pads internally before decoding if it is missing.
What length is rejected?
A length that, after removing whitespace and padding, leaves exactly 1 character in the final group, which cannot be a valid base64 tail.
What happens with an invalid character?
The tool refuses, naming the character and its offset in the cleaned string.
What does the output look like?
Valid UTF-8 text prints directly; anything else, an image or a compressed blob, downloads as decoded.bin.
Can I use this from the command line?
Yes, the same code runs on the page, the CLI and MCP.