There is no mechanism preventing a player from calling refund() in the same block as selectWinner(). A player who identifies they are not the winner (by simulating selectWinner off-chain) can front-run the winner selection with a refund() transaction, exit the raffle for free, and simultaneously manipulate totalAmountCollected inside selectWinner() — reducing the prize and the fee the owner receives.
selectWinner() computes totalAmountCollected = players.length * entranceFee at execution time, but players entries are set to address(0) when a player refunds — they are not removed. The length stays the same, so totalAmountCollected includes the refunded slot, but the ETH has already left the contract.
A front-runner who calls refund() just before selectWinner() confirms:
Gets their entrance fee back
Causes selectWinner() to overestimate totalAmountCollected — it charges the contract for ETH it no longer holds, sending the winner and owner shares that sum to more than the actual balance, or causing a revert
Additionally, if a player can predict they will not win (deterministic randomness based on block.timestamp and msg.sender), they can avoid any loss by front-running to exit before the winner is drawn.
Likelihood:
Requires MEV infrastructure or manual mempool monitoring, which is accessible to any sophisticated actor. The deterministic randomness (block.timestamp + msg.sender) makes winner prediction trivial.
Impact:
A losing player avoids their loss at no cost. The owner's fee is reduced or the entire selectWinner() call reverts, requiring all remaining players to wait for another draw or call refund() themselves — effectively a griefing DoS on the raffle completion.
Alice predicts she will not win by simulating selectWinner() with the current block parameters. She front-runs with refund(), recovers her fee, and the subsequent selectWinner() call tries to transfer ETH it no longer holds.
The contract attempts to distribute more ETH than it holds, confirming the front-running attack breaks winner selection.
Add a time lock that prevents refund() after raffleStartTime + raffleDuration, and compute totalAmountCollected from the actual contract balance rather than players.length * entranceFee:
## Description Malicious actors can watch any `selectWinner` transaction and front-run it with a transaction that calls `refund` to avoid participating in the raffle if he/she is not the winner or even to steal the owner fess utilizing the current calculation of the `totalAmountCollected` variable in the `selectWinner` function. ## Vulnerability Details The PuppyRaffle smart contract is vulnerable to potential front-running attacks in both the `selectWinner` and `refund` functions. Malicious actors can monitor transactions involving the `selectWinner` function and front-run them by submitting a transaction calling the `refund` function just before or after the `selectWinner` transaction. This malicious behavior can be leveraged to exploit the raffle in various ways. Specifically, attackers can: 1. **Attempt to Avoid Participation:** If the attacker is not the intended winner, they can call the `refund` function before the legitimate winner is selected. This refunds the attacker's entrance fee, allowing them to avoid participating in the raffle and effectively nullifying their loss. 2. **Steal Owner Fees:** Exploiting the current calculation of the `totalAmountCollected` variable in the `selectWinner` function, attackers can execute a front-running transaction, manipulating the prize pool to favor themselves. This can result in the attacker claiming more funds than intended, potentially stealing the owner's fees (`totalFees`). ## Impact - **Medium:** The potential front-running attack might lead to undesirable outcomes, including avoiding participation in the raffle and stealing the owner's fees (`totalFees`). These actions can result in significant financial losses and unfair manipulation of the contract. ## Recommendations To mitigate the potential front-running attacks and enhance the security of the PuppyRaffle contract, consider the following recommendations: - Implement Transaction ordering dependence (TOD) to prevent front-running attacks. This can be achieved by applying time locks in which participants can only call the `refund` function after a certain period of time has passed since the `selectWinner` function was called. This would prevent attackers from front-running the `selectWinner` function and calling the `refund` function before the legitimate winner is selected.
The contest is live. Earn rewards by submitting a finding.
Submissions are being reviewed by our AI judge. Results will be available in a few minutes.
View all submissionsThe contest is complete and the rewards are being distributed.