Keys, wallets & account abstraction

MPC wallet

An MPC wallet replaces the single private key with several mathematical shares spread across different parties or devices, and uses secure multi-party computation to sign transactions jointly — so that the full key is never assembled in one place at any moment, not even during signing. The whole-key never exists on a single laptop, phone, or server, which removes the classic single point of theft.

Technically this is built on threshold signature schemes (TSS): protocols like GG18/GG20 for ECDSA or FROST for Schnorr let, say, 2 of 3 share-holders run an interactive computation that outputs one perfectly ordinary signature. The key insight is that the shares jointly compute a valid secp256k1 or Ed25519 signature without ever reconstructing the secret scalar. Many implementations also support proactive resharing, periodically refreshing the shares so that an attacker must compromise enough of them within the same epoch.

It is important not to confuse an MPC wallet with an on-chain multisig. A multisig produces several distinct signatures that a smart contract or Bitcoin script verifies, so the m-of-n policy is visible on-chain, chain-specific, and costs extra gas. An MPC wallet produces one signature from one externally-normal-looking key, so it works on any chain, leaves no special on-chain footprint, and is cheaper to verify — but the policy logic lives off-chain in the signing protocol, which means its correctness, availability, and the security of the software running it become the things you must trust, rather than an audited on-chain contract.

MPC and multisig solve the same goal — no single point of failure — at different layers. MPC hides the policy inside cryptography (one on-chain signature, off-chain trust in the protocol); multisig exposes it on-chain (many signatures, trust in an audited contract). Neither is strictly safer; they shift where the risk lives.

Also called
TSS walletthreshold-signature wallet門檻簽章錢包