Flow

Github

https://github.com/Cyfrin/2024-10-sablier/blob/8a2eac7a916080f2022527408b004578b21c51d0/src/abstracts/SablierFlowBase.sol#L262

Summary

Although the protocol intentionally applies new fees retroactively to all pending withdrawals as a business strategy, this mechanism can be circumvented through front-running by attackers. Users monitoring the mempool can detect fee increase transactions and quickly withdraw their funds before the new fee takes effect.

Impact

MEDIUM - While no funds are at risk, this edge case:

• Reduces protocol revenue

• Creates unfair advantage for tech-savvy users

• Makes fee changes unpredictable

• Undermines intended business model, so basically protocol loses the expected fees.

What causes it?

The fee is determined at withdrawal time rather than when tokens were streamed, making it vulnerable to front-running when fees are increased.

Likelihood

LOW-MED - This attack is:

• Easy to execute

• Highly profitable

• Requires only mempool monitoring

• Can be automated

• Most likely during significant fee increases which will be not frequent ( so low likelihood )

Proof of Concept

front-running Scenario ( where user avoids high fees ):

1. Admin: Let say initial fee is set to 0 % and then admin changed it to setProtocolFee(token, 10%), now transaction is in mempool

2. User: Spots this transaction in mempool

3. User: Quickly submits withdrawal with higher gas

   • withdraw(streamId, fullAmount) // Front-runs with higher gas

4. Admin's fee change executes after

5. User gets funds with original lower fee (0%)

Recommendation

There can be many solutions for mitigating these kinda frontrunning issues, the one I have in mind is to use Flashbots or any private memepool service for updating fees like this.