FAQ
Why not perform threat detection at the smart contract-level?
The Safe Sequencer (custom execution client) can view the whole call stack and interpret each step of execution whereas smart contracts have restricted access to information. We run exploit detection models in the Safe Sequencer that simply would not be possible in a smart contract.
Some alternative solutions hook into the smart contract-level and use external nodes to validate transactions that are subsequently sent to the real sequencer. This approach requires smart contract modifications and relies on simulation results that can easily break, effectively turning the external nodes into slow sequencers of a single application without the cost reduction of dedicated blockspace.