Nova Shield: pre-sign security engine that simulates every request locally and screens it against signed threat lists — unlimited approvals, drainers, honeypots, and lookalike poisoning flagged in plain English.
Every crypto wallet asks you to be the security engine: read the request, spot the trap, judge the hex. Nova Shield moves that check into the wallet. Before you sign anything, Shield simulates the outcome locally and screens the request against cryptographically signed threat lists — then shows you a plain-English verdict. You still decide. You just decide with the outcome in front of you instead of a prayer.
| Exploit class | What the attacker shows you | What Shield shows you |
|---|---|---|
| Unlimited approvals | "stake to earn" | "UNLIMITED USDT spend — 0x…071, signed drainer list" |
| Permit2 signatures | "verify ownership" | "if signed: −1,240 USDT · −0.8 ETH" |
| setApprovalForAll traps | "claim your NFT" | "full collection approval → contract 0x…" |
| Known drainer addresses | a fresh-looking site | "signer on drainer list v214" |
| Honeypot bytecode | "free mint" | "honeypot pattern flagged" |
| Lookalike poisoning | a familiar address | "first 4 / last 4 match — rest differs" |
| Blind eth_sign | a hash you can't read | the decoded contents |
| Clipboard & dust bait | your own history | "copy from the address book, not history" |
Action: Approve · unlimited USDT spend
Spender: 0x6a5a…071 · signed drainer list
If signed: −1,240.00 USDT · −0.8 ETH
Basis: local simulation + threat list v214
Evaluated on-device in 0.42s. Nothing sent.
That's the pattern behind the $494M Scam Sniffer counted stolen to drainers in 2024 — one signature at a time. The full teardown: how wallet drainers actually work.
Shield runs the request against a local fork of state and computes what happens if you sign. "It might be fine" becomes "if signed: −1,240 USDT." That's the number that matters.
Threat lists ship cryptographically signed and evaluate on-device. There's no remote service deciding what you may sign — the lists are data you can verify, and the evaluation never leaves your machine.
Shield sits at the last step of a chain built on absence: phrases entered in an air-gapped input context (never in the DOM, clipboard, or a network path), keys scoped per-chain in isolated keystores, secrets held only in locked, zeroed buffers, and zero telemetry in the binary. Attack anatomy — where the kill happens: the security page, stage 4.
A pre-sign security engine built into the wallet: it simulates the outcome of every external request locally and screens it against signed threat lists, showing a plain-English verdict before you sign.
No. Shield evaluates requests — it never touches key material. Phrases live in an isolated input context and secrets in locked, zeroed buffers; there is no code path between Shield and your keys.
You still decide — Shield shows the verdict, not a lock. Its job is to make the outcome visible ("if signed: −1,240 USDT") and the pattern known (drainer list, honeypot, lookalike) so your decision is informed, not blind.
Threat lists ship signed and evaluate on-device; local simulation runs against a fork of state. There is no remote kill switch deciding what you may sign.
Directly: signature drainers are the #1 airdrop scam pattern (a 'claim' that's really an unlimited approval). Shield flags exactly that pattern — see the airdrop safety checklist.
No — it's local and pre-sign, by design. It runs on-device in ~0.4s; nothing is sent anywhere.