Real attacks you can run yourself. Each one names the layer that stopped it.
- Stolen classical key 3
- Stolen agent key 3
- Forged, tampered or replayed 5
- Fooling the AI 2
On QoreChain a stolen classical key can only do what your post-quantum key has left open. Fenestella shows you that opening, cuts it down to the contracts, amount and time you need, and gives dApps and agents one way to ask for it.
Reading an account needs no wallet and no gas. Works on testnet and mainnet.
▶ Watch the 2:55 demoRead from the chain and from the real test runs each time this page refreshes. Swipe, or let it play.
Real attacks you can run yourself. Each one names the layer that stopped it.
Fenestella reads the chain and tells you which, in a sentence you can act on. If something could not be read, it says so instead of guessing.
Post-quantum key registered, no window open. A stolen classical key moves nothing.
A window for named contracts or plain transfers only, with a count, a cap and an expiry.
A window with no scope. Any contract, any recipient, until it runs out.
No post-quantum key yet. Whoever holds the classical key can register theirs first.
Post-quantum key, proof of possession, encryption key, live window, linked keys and their limits, read straight from the chain.
See a live account →A window limited to named contracts or to plain transfers. Your wallet signs it with the hybrid ML-DSA-87 signature; the chain enforces every bound.
How windows work →One call, requestWindow(). The app states what it needs, you see its address and the bounds in plain words, and you approve in your own wallet.
We play the attacker ourselves, and you can too. In the attack lab every attempt runs live on testnet and is refused by the chain, the vault, the seal or the AI, with a receipt when a transaction reached the chain.