This scheme is a research preview. We want you to find weaknesses.

If you have found a distinguisher, key recovery, or other attack against Dimensional Encryption, please either:

1. Open a public issue on this repository describing the attack, or
2. Contact the author privately at ali@silentbot.com if you prefer coordinated disclosure before going public.

Both are welcome. There is no bug bounty, but we will credit you prominently in any revision of the paper or scheme.

• A distinguisher that beats random guessing with non-negligible advantage, using fewer resources than brute force.
• A key recovery algorithm that succeeds with non-negligible probability using fewer queries or less time than brute force.
• A forgery attack against the authenticated mode (DE-CTR-HMAC).
• A practical side-channel attack against the reference or C implementation (timing, cache, power). These are implementation bugs, not scheme issues, but we want to fix them.

If you want to try attacks on a concrete target before committing to analysis of the full scheme, we can provide reduced-parameter instances (smaller block size, fewer layers, weakened families). Open an issue titled "Request: challenge instances" and we will generate some.

• The Python reference implementation is not constant-time. Known and documented.
• The scheme uses HMAC-SHA-256 or BLAKE3 for the hash layer; if either is ever broken as a PRF, the scheme's fault tolerance kicks in, but we still want to know.
• Dimension 5 (EC analog) is individually vulnerable to Shor's algorithm. This is known and documented.