Fletcher-8 checksum
Fletcher-8 checksum. A fast error-detection checksum, not a cryptographic hash. No authoritative spec exists for this word size (unlike -16/-32/-64): this implementation sums 4-bit nibbles mod 15, low nibble then high nibble per byte. A different convention (whole bytes mod 15) gives a different digest. Check your other tool uses the same convention before comparing.
Runs on your device. The file is never uploaded.
Fletcher-8 checksum runs two sums over the 4-bit nibbles of your file, low nibble before high, and prints two hex digits under a line naming the tool and the file. Both sums are taken modulo 15. Only 225 of the 256 two-digit values can ever come out, so collisions arrive on anything longer than a few bytes.
Questions
What is Fletcher-8 for?
Detecting accidental corruption in small payloads, where one byte of checksum is all the space you have. It is an error-detection checksum, not a security hash: anyone can make a different input with the same value.
Why does my other tool give a different result?
Because no authoritative spec exists for this word size, unlike Fletcher-16, 32 and 64. This implementation sums 4-bit nibbles modulo 15, low nibble first, then high. A tool that sums whole bytes modulo 15 produces a different digest. Check the convention before comparing.
How long is the output?
Two hex digits. That is 256 possible values, so collisions are common by design. Use it as a cheap corruption check on a short message, not as an identifier.
Which should I use instead for integrity?
For accidental corruption on real files, CRC32 or xxHash3 are faster and wider. For anything an attacker might touch, use SHA-256.