Detoxer is a suite of custom Uniswap v4 hooks designed to discourage MEV attacks (especially sandwich), punish attackers, Revard victims, protect regular users, and Ensure LP have the same fee collected by dynamically adjusting swap fees. Each hook independently targets a different class of exploit pattern. These hooks can be combined to create a powerful complex hook to fight against MEV attacks.
Deck - Google Slides Video - YouTube
Located in src/, each hook is plug-and-play compatible with Uniswap v4’s Hook system.
-
Goal: Deter price manipulation during the each block.
-
How it works:
- Captures the initial price at the start of each block (on the first swap).
- Calculates how much the price shifts during subsequent swaps within the same block.
- Applies a dynamic fee based on how favorable (or suspicious) the price movement is:
- 📈 2.0% fee — when a trade aggressively pushes the price in a profitable direction (potential attacker).
- ⚖️ 0.9% fee — for moderate or neutral price shifts.
- 📉 0.5% fee — when a trade worsens the price (likely a victim).
-
Why it matters: Attackers often front-run early in a block and than back-run later, both part of sandwich usually have better price than victim's one. By making price-moving trades more expensive, this hook disincentivizes predatory behavior and protects regular users.
-
Goal: Dynamically adjust swap fees based on the transaction’s priority fee (tip) to target MEV-style behavior.
-
How it works:
- Measures the difference between
tx.gaspriceandblock.basefeeto get the priority fee. - Maintains a moving average of recent priority fees.
- If a user is paying higher-than-normal priority fees, it's treated as a signal of MEV or aggressive behavior.
- Swap fee is adjusted proportionally:
- 🔺 High tip = higher fee (up to 5.0%)
- ⚖️ Normal tip = base fee (1.0%)
- 🔻 Low tip = minimum fee (0.5%)
- The priority fee average is updated over time with a smooth exponential moving average (EMA).
- Measures the difference between
-
Why it matters: MEV bots often pay abnormally high tips to get their txs included faster. This hook detects and charges them more, while protecting normal users with lower fees.
Requires the pool to be configured with an initial average priority fee (passed to the constructor).
-
Goal: Detect and penalize sandwich attacks by identifying when a trader swaps in both directions within the same block.
-
How it works:
- Tracks the most recent block in which each user swapped in a specific direction (
zeroForOneoroneForZero). - If a user swaps in the opposite direction in the same block, the hook considers it a sandwich and applies a higher fee.
- Fee logic:
- 🔄 Same-direction swap: charged 0.9% (
BASE_FEE) - 🔁 Both-direction (sandwich) swap: charged 3.6% (
MAX_FEE)- Effectively 1.8% on front-run, 1.8% on back-run.
- 🔄 Same-direction swap: charged 0.9% (
- Tracks the most recent block in which each user swapped in a specific direction (
-
Why it matters: Sandwich bots often front-run and back-run swaps within a single block. This hook flags that pattern and makes it expensive, reducing MEV profits and deterring exploitative behavior.
-
Goal: Penalize transactions that appear early in a block, especially those sent by MEV bots attempting to front-run.
-
How it works:
- Tracks each transaction's position in the current block.
- The first swap in a block pays the maximum fee (1.05%).
- For each new unique
tx.originwithin the same block:- The fee is reduced by 50% relative to the last one (down to a minimum of 0.5%).
- This creates an incentive for users to appear later in the block, and disincentivizes frontrunning.
- If the same
tx.originappears multiple times (potential backrun), it is charged the maximum fee again. - Uses transient storage to distinguish between multiple calls in the same transaction vs different transactions.
-
Why it matters: MEV bots typically try to be first in a block. This hook flips the incentive — making first-in-line trades more expensive, and rewarding users who aren't manipulating the block.
-
Goal: Combine all the above hooks into a single, powerful hook to fight against MEV attacks.
-
Currently not implemented
- All hooks were backtested using Uniswap V3 data from a low-liquidity LAI-USDT pool
- ~1,000 real historical transactions were processed:
- 900+ swaps
- 20+ liquidity events (adds/removes)
- Included analysis of 20+ real sandwich attacks
- The initial testing results are available in the
data/reports/directory.
| Category | Avg. Fee Delta (Hook) | Vanilla Pool | Difference |
|---|---|---|---|
| total | 9,932 | 10,000 | -68 |
| attacker | 15,350 | 10,000 | +5,350 |
| victim | 5,333 | 10,000 | -4,667 |
| regular | 9,681 | 10,000 | -319 |
💡 This shows the hook successfully increases fees for attackers while reducing the burden on victims and regular users.
| Category | Avg. Fee Delta (Hook) | Vanilla Pool | Difference |
|---|---|---|---|
| total | 9,972 | 10,000 | -28 |
| attacker | 19,775 | 10,000 | +9,775 |
| victim | 6,817 | 10,000 | -3,183 |
| regular | 9,372 | 10,000 | -628 |
💡 This hook heavily penalizes attackers while modestly reducing fees for victims and regular users.
| Category | Avg. Fee Delta (Hook) | Vanilla Pool | Difference |
|---|---|---|---|
| total | 10,035 | 10,000 | +35 |
| attacker | 24,750 | 10,000 | +14,750 |
| victim | 9,000 | 10,000 | -1,000 |
| regular | 9,031 | 10,000 | -969 |
💡 This hook strongly penalizes attackers, while also slightly reducing fees for regular users and victims.
| Category | Avg. Fee Delta (Hook) | Vanilla Pool | Difference |
|---|---|---|---|
| total | 10,012 | 10,000 | +12 |
| attacker | 10,408 | 10,000 | +408 |
| victim | 5,250 | 10,000 | -4,750 |
| regular | 10,118 | 10,000 | +118 |
💡 This hook mildly increases fees for attackers and strongly reduces them for victims, while regular users see near-neutral impact.
-
Adjust FEE values inside the hook contracts located in
src/ -
Run the Foundry test suite:
forge test -vv -
Check the results in
data/reports/for each hook- Each hook generates a CSV file with the results of the test
- The CSV files contain the swap count, total fee delta, and average fee delta for each type of swap participant
Each result file contains 4 rows, representing different types of swap participants:
- total — All swaps included in the test
- attacker — Swaps identified as part of sandwich attacks
- victim — Swaps negatively affected by sandwich attacks
- regular — Neutral users not involved in MEV activity
Each row includes 3 values:
- Swap count — Number of swaps in that category
- Total fee delta — Total additional fee paid (or saved) due to the hook
- Average fee delta — Average fee impact per swap
Add new features:
- Techniques:
- Front/back run liquidity modification fee
- Past activity historical karma/rate fee
- Cross-pool activity logic fee
- Arbitrage like behaviour identification
- Front/back run amount comparison fee
- High activity fee etc.
- Integrations:
- Brevis (off-chain mev proofs)
- Chainlink/EigenLayer (fair price feed)
- Arbitrum Stylus (gas efficient Rust implementation)
- FlashBots/Zeromev (MEV API for backtesting)
Improve backtesting:
- Variety:
- Low/high liquidity
- Low/high base fee level
- Low/high volume
- New sandwich patterns
- Arbitrage protection testing
- Measure net profit (not fee level)
- Automation:
- One-click setup for any pool
- UI to visualize results
- Optimized API usage
Optimize solution:
- Gas:
- Storage management
- Efficient Math
- Reuse of data
- Fee:
- Tune to better values
- Compatible with any pool
- Data:
- Emit standard events (OZ)
- Use hookData (improve protection)