XOR Key Recovery
Recover the key from XOR-obfuscated data without being told what it is. If you know how the decrypted file should start, the key falls out immediately.
A short crib still helps: it pins the bytes it covers and frequency analysis fills in the rest.
Intended for your own single-player save files. Editing saves in multiplayer or online games breaks their terms of service and can get your account banned. Always keep a backup of the original file before you change anything.
Repeating-key XOR leaks its key whenever you can guess part of the plaintext — XOR the ciphertext against what you expect and the key falls straight out. The automatic method needs a reasonable amount of data, roughly twenty bytes per key byte, so it works far better on a full save than on a short fragment.
How to use
- 1Load the obfuscated file.
- 2Pick a method — a known opening is by far the most reliable.
- 3For JSON saves, the default crib of a brace and quote is usually enough.
- 4Click Recover, then pick the candidate whose preview reads correctly.
Frequently Asked Questions
How does the known-plaintext method work?
XOR is its own inverse, so ciphertext XOR plaintext gives the key directly. If you know the file starts with an opening brace and quote, XORing those against the first bytes reveals the first bytes of the key. The crib must be at least as long as the key to pin down every byte.
I do not know how the file starts. What then?
Use the automatic method. It estimates the key length from repeating patterns, then picks each key byte by choosing whichever value produces the most plausible text. It needs a decent amount of data — roughly twenty bytes per key byte — so it works well on a full save and badly on a fragment.
Why does it show several candidates?
Because more than one key length can be consistent with the evidence. Each candidate shows a preview of the decrypted start, so you can pick the one that actually reads as text. The confidence figure is a guide, not a guarantee.
Why does frequency analysis fail on short files?
It relies on seeing enough characters per key position to judge which candidate byte produces sensible text. With only a handful of samples per position the statistics are meaningless — and the tool tells you the sample count so you know when to distrust the result.
Does this work on real encryption?
No. These techniques exploit the fact that XOR preserves plaintext structure. AES and other real ciphers destroy it entirely, and no amount of analysis recovers the key without a fundamental break.