Threat Model
DeFi signing threat analysis: clipboard hijacking, blind signing, extension compromise, memory scraping, TX substitution, and approval farming. Why private keys must never exist on networked machines.
Simulation proves a transaction is safe to execute. But safe calldata is worthless if the signing environment is compromised. This page covers the threats; the Signing Flow shows how x0x1 defends against each one.
Why signing on a hot machine is dangerous
A "hot machine" is any computer connected to the internet. This includes your desktop, your phone, and most laptops. The threat surface is enormous:
What Safe Signing protects against
Keys and internet never coexist on the same device. Physical separation, not software isolation:
What Safe Signing does NOT protect against
- Smart contract bugs in the target protocol (mitigated by Anvil simulation + risk checks)
- Social engineering of the user (mitigated by state diff review)
- Physical compromise of the signing device (mitigated by hardware wallet integration)
- Compromised RPC provider returning false state. Anvil fork catches execution-level lies. Multi-RPC cross-check is planned but not yet implemented.
Now that you know the threats, see exactly how the Signing Flow implements physical device separation with a three-device architecture.