Imagine you sell an NFT for 10 ETH to a stranger in another city. Without escrow, one of you takes a risk. Classic services trust a middleman, but in crypto a "trusted" party is either a centralized platform (vulnerable to regulatory blocks) or a smart contract: it never sleeps, takes no bribes, executes code deterministically. We develop turnkey escrow contracts with arbitration, timeouts, and support for any tokens. Over 5 years we have built more than 20 systems with total TVL exceeding $5M, each formally verified. Our contracts are protected against reentrancy, front-running, and flash loan attacks. Get a free consultation—we'll evaluate your project at no cost.
Why an Escrow Contract Is Safer Than an Exchange Deposit
Centralized exchanges and escrow services hold your funds on their wallets. In case of a hack or regulatory freeze, money gets locked. A smart contract has no human factor: the code is open, audited, and immutable after deployment. An audited smart contract reduces fraud risk by 5 times compared to a centralized escrow service. The cost of developing such a contract is determined after analysis—typically a fraction of the savings on platform fees.
Basic Mechanics and Where They Break
Simple escrow: buyer deposits funds → seller fulfills condition → buyer confirms → funds released. Problem: if the buyer never confirms, funds are locked forever. A minimal correct scheme requires:
- Timeout with automatic refund—if the buyer hasn't confirmed within N days, the seller can request a return. Or vice versa: if the seller fails to deliver, the buyer withdraws the deposit.
- Arbitration—a third party with override power: a specific arbitrator (address), multisig, or DAO.
- Dispersion model—the arbitrator doesn't hold the funds but decides a split (e.g., 75% buyer, 25% seller).
How to Prevent Arbitrator Collusion?
The hardest part is not the mechanics but the arbitration model. If the arbitrator has absolute power, they become an attack target (bribe, key compromise). If chosen by the parties, there is collusion risk. Practical patterns:
- Commit-reveal arbitration. Both parties send encrypted decisions; the arbitrator reveals theirs only after receiving both. Doesn't eliminate collusion but complicates it.
- Claim-opponent arbitration (ERC-792 style). Each party provides document hashes; the arbitrator votes publicly, the ruling is recorded on-chain and auditable.
- Random arbitrator from a pool. Kleros Protocol uses a decentralized court—random selection of jurors from stakers, economic incentive to vote honestly. We integrate via
IArbitrable/IArbitratorinterfaces.
Contract structure with Kleros support:
contract Escrow is IArbitrable {
IArbitrator public immutable arbitrator;
uint256 public disputeId;
enum Status { Pending, Active, Disputed, Resolved }
struct Deal {
address buyer;
address seller;
uint256 amount;
uint256 timeout;
Status status;
uint8 buyerPercent;
}
function raiseDispute(uint256 dealId) external payable {
Deal storage deal = deals[dealId];
require(deal.status == Status.Active);
require(msg.value >= arbitrator.arbitrationCost(""));
deal.status = Status.Disputed;
disputeId = arbitrator.createDispute{value: msg.value}(
2,
""
);
emit Dispute(arbitrator, disputeId, dealId);
}
function rule(uint256 _disputeId, uint256 _ruling)
external override
{
require(msg.sender == address(arbitrator));
_executeRuling(_disputeId, _ruling);
}
}
How to Choose an Arbitration Model?
| Model | Speed | Collusion Resistance | Cost |
|---|---|---|---|
| Single arbitrator | <1 day | Low | Free |
| Multisig (3 of 5) | 1-3 days | Medium | Gas |
| Kleros (random) | 2-7 days | High | Small stake |
Gas optimization tip
Use commit-reveal arbitration only for high-value deals: each voting round costs gas. For deals under $1K, multisig is sufficient.Why Timeout Is Critical
Without a timeout, funds are locked forever if one party goes silent. Typical timeouts: 3 to 30 days. On timeout, the seller requests a refund (if buyer didn't confirm) or the buyer withdraws (if seller didn't deliver). Timeout pauses during an arbitration dispute.
ERC-20 vs Native ETH: Non-Obvious Differences
Escrow with ETH is simpler—send msg.value, return via call. With ERC-20, approve is needed before deposit. This creates two attacks:
- Token approval front-running. Classic: user approves(spender, 100), then approves(spender, 200). In between, attacker withdraws 100. Solution: always approve(spender, 0) before new approval, or use
permit(EIP-2612). - Fee-on-transfer tokens. Deflationary tokens charge a fee on transfer. Need to check actual received:
uint256 before = token.balanceOf(address(this)); token.transferFrom(...); uint256 received = token.balanceOf(address(this)) - before;
For fee-on-transfer tokens, we use SafeERC20 from OpenZeppelin, which correctly handles non-standard tokens (e.g., USDT doesn't return bool). OpenZeppelin SafeERC20 is available on GitHub.
Multi-Token Escrow
To support both ETH and ERC-20, unify the interface via the zero address for ETH:
function deposit(address token, uint256 amount) external payable {
if (token == address(0)) {
require(msg.value == amount);
} else {
require(msg.value == 0);
IERC20(token).safeTransferFrom(msg.sender, address(this), amount);
}
}
Comparison with Centralized Platforms
| Parameter | Centralized Platform | Smart Contract | Our Hybrid Approach |
|---|---|---|---|
| Trust | Full trust in platform | Trust in code | Trust in audited code + arbitration |
| Fee | 1-3% of deal | ≈0% (gas only) | 0% for arbitration, gas optimized |
| Regulatory risk | High (account freeze) | Low | Low |
| Transaction speed | Instant | 5-10 blocks | Optimized to 5 blocks |
How to Integrate Kleros Arbitration
- Deploy an Escrow contract inheriting
IArbitrable. - Specify the
arbitratoraddress in the constructor. - In
raiseDispute, callarbitrator.createDisputewith evidence. - Implement
ruleto process the arbitrator's ruling. - Run fuzzing tests with Foundry—cover edge cases of timeouts and fee-on-transfer.
Development Process
- Design (0.5-1 day). Define: arbitrator, timeouts, tokens, partial release.
-
Development and tests (2-4 days). Foundry with fuzzing—boundary cases for timeouts, fee-on-transfer, recursion. Separate reentrancy tests using
ReentrancyGuard. - Audit and deployment. For TVL > $100K, external audit (1-2 weeks). Deploy with verification on Etherscan/Polygonscan.
The average development cost varies depending on complexity; start with a free consultation. Contact us for a detailed discussion of your scenario. Order escrow contract development with audit—get reliable protection for your deals.