Not because the cryptography is unusual. It is standard, and that is the point. The differences are in what gets told to you.
The proving system here is not novel and it should not be. Fixed denominations, Poseidon commitments, a Merkle tree and a Groth16 proof are a well-understood design with a decade of scrutiny behind them. Novelty in this particular corner of software is a warning sign, not a feature.
Every pool has two anonymity-set figures: how many deposits it holds, and how many of those the operator does not already own the notes for. The second is smaller and less flattering. Both are shown. A tool that quotes only the first is describing a pool that hides you from strangers while its operator reads over your shoulder.
A brand-new destination wallet holds no SOL and cannot pay its own fee. Funding it from your old wallet rebuilds precisely the link you just paid to cut. So a relayer broadcasts instead and takes its margin from the withdrawal — and because the recipient and fee are bound into the proof, it cannot redirect or overcharge. The destination ends with one transaction in its history and no visible funding source.
The proving key came from a multi-party ceremony where each contributor folds in randomness and publishes a fingerprint. The chain of fingerprints is on the Proof & trust tab with its current status stated plainly, including where it is incomplete. An unpublished contribution is the operator's word, and it is labelled as such rather than counted.
Not "100% anonymous". Not "untraceable". Not "no logs" — the server sees requests and the page loads analytics, and that is spelled out. What is claimed is narrow and testable: the deposit and the withdrawal cannot be linked on chain, and how well that protects you depends on the size of the pool and how long you wait.
Compare: against Tornado Cash, or against an exchange.