System Architecture.
Architecture is not about what you build. It is about what you remove.
Most scaling solutions are just digital banks run by committees. They call it a bridge, it is a multisig. They trade the certainty of code for the whims of signers.
Ztarknet optimizes the human out of the loop. There are no admins, no federations, and no "honest majority" assumptions.
There is only a Transparent Zcash Extension (TZE) enforced by consensus. The system does not ask for permission. It asks for a proof.
The only authority here is mathematics.
The Bridge (The TZE)
- The Mechanism: A "Transparent Zcash Extension" (TZE). Think of it as a portal in the Zcash code that lets it read STARK proofs.
- The Guarantee: It is not a multisig. It is a Consensus Rule. If the L2 tries to steal funds, the L1 math rejects the block.
The Eyes (Zoro)
- What it is: A "Light Client" that verifies Zcash.
- The Magic: It shrinks the entire history of Zcash (gigabytes) into a 50KB receipt. This lets mobile phones and the L2 verify Zcash state without trusting a server.
The Infrastructure (Karnot)
- Who they are: The engine room. The team keeping the lights on.
- What they provide: The physics for the math. RPC nodes. Explorers. Invisible rails.
The Interface (Session Keys)
- The Bottleneck: Zcash wallets cannot sign game moves yet. Manual signing kills the speed.
- The Fix: A temporary browser key. It signs interactions automatically. You approve once. You play infinitely. Web 2 UX.
The Stack
- Sequencer: Madara (Rust-based Starknet Client).
- Prover: Stwo (Circle-STARKs).
- L1 Node: Zebra (Zcash Node with TZE fork).
- Explorer: Cosmos (Open-source Starknet explorer).
docker compose -f docker-compose.devnet.yml up
User Transaction
Madara
Sequencer
Generate Trace
Generate Trace
Stwo
Prover
Proving
Generate ZK Proof
Zebra
L1 Node
Verification
TZE Verifier
L1 Anchor
State Updated