Cross-Chain Bridge & Interoperability Development

Cross-Chain Bridge Development: Architecture, Risks, and Implementation

We develop cross-chain bridges and cross-chain solutions end-to-end. We know how to avoid disasters. A few years ago, the Binance BNB Chain bridge lost $570M — the attacker forged a Merkle proof in BSC's native bridge. That same year, Wormhole lost $320M: guardian signature verification was bypassed through a bug in Solana's secp256k1 program. Ronin Bridge — $625M. These are not coincidences. Bridges are the most attacked infrastructure in Web3 because they aggregate liquidity and have complex cross-chain verification logic.

Why Do Bridges Break? Three Architectural Classes of Vulnerabilities

Finality and Reorg Issues. Ethereum has probabilistic finality before The Merge and economic finality after (2 epochs, ~12 minutes). Bitcoin — ~6 blocks (~60 minutes). Solana — ~400ms. If a bridge mints wrapped tokens on the destination chain immediately after 1-2 blocks on the source — a reorg of 3+ blocks allows the attacker to obtain tokens on the destination while the source transaction is reverted. Correct protection: wait for finality confirmation specific to each chain. For Ethereum — 64+ blocks (2 epochs). Not one block.

Signature Verification. Most bridges use a multisig committee or threshold signature: N out of M validators must sign the event from the source chain. Wormhole used 13 out of 19 guardians. The attack was not on the keys themselves — the attacker found a vulnerability in the signature verification code on Solana, where an outdated sysvar account was accepted as valid without verification. On-chain signature verification is harder than it seems.

Lock-and-Mint vs Burn-and-Mint. In the lock-and-mint model, original tokens are locked in a contract on the source chain, and wrapped tokens are minted on the destination. The source contract is a honeypot: all locked TVL is there. One bug in the unlock logic — and all funds are available to the attacker without needing to do anything on the destination chain. Native burn-and-mint (like Circle CCTP for USDC) is safer: no locked pool.

How to Choose a Messaging Layer for Your Project?

LayerZero — a protocol for arbitrary message passing between chains. Not a bridge itself, but infrastructure for building bridges and omnichain applications.

Architecture: Endpoint contract on each chain, Executor (delivers messages to the destination chain), DVN (Decentralized Verifier Network — verifies the transaction fact on the source chain).

Source chain:
  OApp.send() → Endpoint.send() → [emits packet event]

Destination chain:
  DVN verifies packet hash → Executor calls Endpoint.deliver() → OApp.lzReceive()

In v2, the developer chooses DVNs: official (LayerZero Labs, Google Cloud, Polyhedra), or custom. One can configure required DVN + optional DVN: a message is accepted only if all required DVNs confirm. This allows building bridges with different trade-offs between security and speed.

OApp (Omnichain Application) — the base contract for integration. Inherit OApp, implement _lzSend and _lzReceive. For token bridges — OFT (Omnichain Fungible Token) standard out of the box does burn-on-source / mint-on-destination.

Wormhole uses a network of 19 guardians (large companies like Jump Crypto, Everstake, etc.), each signing observed events. Threshold — 13 out of 19. VAA (Verified Action Approval) — a signed message that is accepted on the destination chain.

Main difference from LayerZero: Wormhole has native support for non-EVM chains: Solana, Aptos, Sui, Algorand, Near. For projects needing a bridge between Ethereum and Solana — Wormhole is often the only production-ready option.

After the exploit, Wormhole added Native Token Transfers (NTT) — an architecture without a locked pool, similar to CCTP. NTT + Hub-and-Spoke model: redundant liquidity is not accumulated on one chain.

Relay Architecture and Light Client Verification

Relay-based bridges (IBC in Cosmos ecosystem, Succinct's Telepathy) verify the source chain's state via a light client on the destination chain. For EVM→EVM: a contract on Ethereum stores and verifies BLS signatures of the source chain's blocks.

ZK-bridges are the next level. Succinct, Polyhedra zkBridge, Electron Labs generate a ZK-proof of the correctness of the source chain's consensus. On the destination chain, the proof is verified, not the validator signatures. Removes trust in the committee. But ZK-proof verification is gas-expensive — from 200k to 500k gas on Ethereum L1 depending on the proof system. A ZK-bridge is safer than a relay-based bridge but requires 2-3 times more gas for verification.

What Does Cross-Chain Bridge Development Include?

We implement the project turnkey and deliver a complete set of results. Our clients receive:

Implementation: What to Consider Before the First Line of Code

Mandatory components for any production bridge:

Pauser. Emergency pause function, called by multisig or automatically upon anomaly detection (suspicious volume, atypical call sequence). Most hacked bridges did not have or did not use a pauser in time.

Rate limiting. Limit output volume per time interval. If an attacker drains the bridge — rate limit gives time to react. Implementation: transferVolume[currentEpoch] += amount; require(transferVolume[currentEpoch] <= epochLimit).

Finality checks. Specific to each chain. Not "wait 1 block", but use finality API or wait for required number of confirmations.

Relayer monitoring. An autonomous service that monitors the state of both bridge sides. If a message is sent but not delivered within N minutes — alert. If locked balance diverges from totalSupply of wrapped token — critical alert.

Timeline and Cost

A simple ERC-20 bridge on top of an existing messaging layer (LayerZero OFT or Wormhole NTT) — 4-8 weeks including testing and audit. A custom bridge with own verification, multi-chain support, rate limiting, monitoring — 12-24 weeks. A ZK-bridge with custom proof circuits — from 6 months.

Bridge audit takes longer than a standard DeFi protocol audit: cross-chain scenarios, finality edge cases, reorg attacks must be tested. Minimum 3-4 weeks for a production-grade solution.

Cost is calculated individually after workload assessment. We have been working since 2018 and have completed 15+ projects in blockchain infrastructure. Contact us — we will evaluate your project and propose the optimal bridge architecture.