Skip to main content
🔨 Planned for T3. Audit-gated mainnet deployment of the full T1 + T2 contract suite, plus mainnet-only operational primitives (timelock, fees, multisig handoffs).
Phase 4 is the launched mainnet architecture. The bulk of the contract work is already in Phases 1–3; Phase 4 hardens the operational surface so the protocol can hold real value.

Architecture diagram

Phase 4 architecture — Mainnet end-state

Delta from Phase 3

What changes vs Phase 3:

Multisig admin + Timelock

Single-key admin replaced by an Admin Multisig on each chain (Stellar account-level multi-sig, Gnosis Safe on EVM). Admin operations flow through a new Timelock module with a 24-hour delay. Pause / unpause exempt for emergency response.

Arbiter Multisig

ArbiterRole moves from ”= admin” to its own dedicated 5-of-7 multisig (3 team + 4 ecosystem). Quorum of 5 from a 3+4 pool guarantees ≥ 2 non-team signers in any decision by construction. Ecosystem signers are sourced via the Stellar Foundation.

Protocol fee on `unlock`

protocol_fee_bps = 20 (0.2%) and lp_fee_bps = 30 (0.3%), capped at 100 bps each. Bridger nets 99.5% of the destination amount; 0.3% stays in the maker’s ad.balance as LP profit; 0.2% routes to the new Fee Pool.

Per-route bond config

BondConfig { bond_token, min_bond, bond_bps } per route. USDC route launches with bond in USDC, floor of 1 USDC, ratio of 50 bps (0.5%). Same-token denomination keeps the math unit-clean without an oracle.

Reconciliation listener

Off-chain HA service, two-region. Watches both chains via private RPC providers, detects settlement discrepancies within a 5-minute SLA, alerts on-call. Read-only — no contract authority.

Bootstrap mode

Initial deploy runs without delay so the admin can wire routes, fees, and roles. activate_timelock() is a one-shot irreversible flag flip before multisig handoff.

Fee math

The fee mirrors the protocol’s published 0.5% total bridge fee, split between the LP and the protocol. It’s deducted destination-side from the maker’s locked amount:
PartySource sideDestination sideNet per trade (1:1 cross-chain)
BridgerDeposits XReceives 0.995 × X-0.5% × X
MakerReceives XLocked balance reduces by 0.997 × X+0.3% × X
Protocol+0.002 × X to fee pool+0.2% × X
Bps values are admin-configurable but timelocked, and capped at 100 bps each in the contract to bound future admin-key abuse.

The launch route — USDC ↔ USDC

ProofBridge does not issue tokens. T3 launches with one production route between two existing token contracts:
SideAssetIssuer
Ethereum mainnetUSDC (ERC-20)Circle
Stellar mainnetUSDC (Stellar Asset Contract)Circle
ProofBridge configures the route via set_token_route on each chain (timelocked); Circle’s contracts are unmodified. Future routes — including any native-token bridging (XLM, ETH) — wait for ecosystem partners to issue equivalents on the counterpart chain. Adding a route post-launch is config-only: set_token_route + set_settlement_window + set_bond_config, all timelocked admin calls.

Mainnet deployment is audit-gated

The SCF Audit Bank engagement runs during T3 and gates mainnet deployment:
1

T2 contract code freezes

End of Phase 3 / Tranche 2. T3 contract changes (timelock, fee mechanism, per-route bond config) are staged but not yet deployed.
2

Audit kickoff

Auditors review the full T1 + T2 + T3 contract suite.
3

Critical / high findings closed

Audit Bank confirms remediation closure.
4

Mainnet deployment begins

Deploy contracts, run initial config, activate timelock, hand off admin to multisig.
5

Production routes verified

First-tx ceremony for the USDC route in both directions. Tx hashes recorded on /reference/smart-contracts.
6

4-week post-launch hardening

Daily uptime reports, audit medium/low remediation, integration partner support, perf tuning.
If audit timing slips, mainnet deploy slips. We don’t ship a known-broken launch.

Trust model at mainnet

For the full per-phase trust comparison see the security model. At mainnet the residual assumptions are bounded by:
  • Multisig + 24h timelock on every admin operation that can move funds, redirect routes, or change auth gating.
  • Separate 5-of-7 arbiter multisig with ≥ 2 non-team signers guaranteed in every decision. We engage the Stellar Foundation to source ecosystem signers for the multisig.
  • Cryptographic gates (BLS aggregate + UltraHonk + nullifier + EIP-712 binding) unchanged from earlier phases.
The arbiter remains the largest residual trust assumption — Phase 5 addresses it by handing the role to DAO governance.

What ships next

Phase 5: Decentralization & Expansion

DAO governance for the ArbiterRole, agent stake-and-slash for misbehavior, additional chains and RWA routes via ecosystem partnerships, and protocol-level optimizations like per-route fee bps and oracle-driven cross-token bonds.