CoW Swap:
genuinely sophisticated MEV protection, with a core protocol that's held even when the edges haven't.
We tore apart CoW Swap, the trading interface for CoW Protocol, across the same aggregator-adapted scorecard we've used throughout this series. From a genuinely distinctive Coincidence of Wants peer-to-peer matching mechanism, letting two traders settle directly against each other with zero AMM fees or slippage on the matched portion, a genuinely sophisticated CoW AMM design that redistributes rebalancing surplus back to liquidity providers instead of letting MEV bots capture it, and gasless, fee-only-on-success order signing, to two real, confirmed incidents worth presenting with precision: a February 2023 solver exploit that drained roughly $166,000-$181,600 in protocol fees, not user funds, with the responsible solver's bond covering the damages, and an April 2025 domain-hijacking attack that redirected users to a phishing site for roughly $1.2 million in losses, while multiple sources confirm the core on-chain protocol itself remained secure in both cases; and landed on a score the marketing page won't show you.
Our take, up front: CoW Swap is the trading interface for CoW Protocol, an intent-based aggregation system genuinely different from a typical router: rather than submitting a transaction directly, users sign an intent, orders are collected off-chain and grouped into batch auctions, and competing solvers propose the best execution. Real, genuinely distinctive Coincidence of Wants matching: when two traders' orders can settle directly against each other, they do, bypassing external liquidity entirely and avoiding AMM fees, price impact, and slippage on the matched portion. Real, genuinely sophisticated CoW AMM design, capturing rebalancing and arbitrage surplus that would normally go to MEV bots and redistributing it to liquidity providers instead, with early results showing meaningfully higher returns than equivalent Uniswap V2-style pools. Real, genuinely thoughtful, user-friendly mechanics: gasless order signing with no ETH required upfront, fees charged only on successful execution, and free order cancellations. Real, genuinely non-custodial architecture, with a real, functioning solver-bond safeguard requiring solvers to post collateral that covers damages if something goes wrong. What we can't set aside: two real, confirmed incidents, though in both cases multiple independent sources confirm the core on-chain protocol itself was never breached. On February 7, 2023, a malicious solver exploited a bug in an intermediary contract to drain roughly $166,000-$181,600 in protocol-collected fees; critically, user funds were never at risk since CoW Swap doesn't hold them, and the responsible solver's bond covered the damages. On April 14, 2025, attackers used social engineering to hijack CoW Swap's domain registration, redirecting users to a phishing site for roughly $1.2 million in losses; CoW Swap's on-chain contracts, backend APIs, and solver networks all remained confirmed secure throughout. Real, honest, significant scope limitation relative to other aggregators in this series: CoW Swap operates on Ethereum and Gnosis Chain specifically, a genuinely narrower footprint than most platforms we've reviewed. We weighted all of it below.
Real, genuinely resilient core protocol: no confirmed breach of CoW Protocol's actual on-chain settlement logic across either incident we found. Real, confirmed incident: on February 7, 2023, a malicious solver exploited a bug in an intermediary contract to drain roughly $166,000-$181,600 in protocol-collected fees; critically, user funds were never at risk since CoW Swap doesn't hold them, and the responsible solver's bond covered the damages. Real, confirmed incident: on April 14, 2025, attackers used social engineering to hijack CoW Swap's domain registration, redirecting users to a phishing site for roughly $1.2 million in losses; multiple sources confirm the core on-chain contracts, backend APIs, and solver networks all remained secure throughout. Real, at least one additional source suggests a further frontend-flagging incident around April 2026, though we couldn't fully confirm whether this is a distinct event or a delayed report of the 2025 incident.
Pros
- Core on-chain protocol confirmed unbreached across both known incidents
- Solver-bond safeguard functioned as designed, covering 2023 damages
Cons
- Feb 2023: ~$166-182K solver exploit (protocol fees, not user funds)
- Apr 2025: ~$1.2M user losses via a domain-hijacking phishing attack
Real, genuinely distinctive, structurally unique execution model: Coincidence of Wants peer-to-peer matching avoids AMM fees, price impact, and slippage entirely on the matched portion of a trade, with batch auctions and solver competition handling the remainder. Real, genuinely honest, well-corroborated limitation: multiple detailed sources note this model is less suited to small, simple swaps in already-deep pairs, where a direct AMM swap may be faster and simpler.
Pros
- Coincidence of Wants matching genuinely eliminates fees/slippage on matched portions
- Batch auctions and solver competition provide real, structural MEV protection
Cons
- Multiple sources note it's less ideal for small, simple swaps in deep pairs
Real, genuinely non-custodial; funds remain in a user's wallet until a solver fills the order on-chain. Real, functioning CoW DAO governance. Real, genuinely positive, disclosed incident-response mechanism: solver bonds specifically exist to cover damages from a compromised or malicious solver, a real structural safeguard that functioned as intended in the 2023 incident.
Pros
- Genuinely non-custodial; functioning CoW DAO governance
- Solver bonds provided real, disclosed compensation when actually needed
Cons
- Solvers themselves represent a real, structural point of potential compromise
Real, honest, significant scope limitation relative to other aggregators in this series: CoW Swap operates on Ethereum and Gnosis Chain specifically, a genuinely narrower footprint than 1inch's 13+ chains or Rango's 50+.
Pros
- Deep, focused aggregation across major DEXs within its supported chains
Cons
- Limited to Ethereum and Gnosis Chain, genuinely narrow relative to this series' other aggregators
Real, genuinely distinctive gasless order signing (no ETH required), fee-only-on-successful-execution, and free order cancellations. Real, disclosed price impact warnings and clear fee breakdowns. Real, honest, well-corroborated limitation: less ideal for users needing guaranteed immediate execution.
Pros
- Gasless order signing; fees charged only on successful execution
- Free order cancellations; disclosed price impact warnings
Cons
- Less ideal for users needing guaranteed, instant execution
Real, genuinely distinctive CoW AMM design that captures rebalancing and arbitrage surplus and redistributes it to liquidity providers rather than letting MEV bots extract it, with early results showing meaningfully higher LP returns than equivalent Uniswap V2-style pools. Real, gas costs bundled into the protocol fee only upon successful execution.
Pros
- CoW AMM redistributes MEV-adjacent surplus to LPs instead of bots
- Success-only fees; free cancellations if a trade never executes
Cons
- None significant found in our research
Real, genuinely distinctive Coincidence of Wants peer-to-peer matching. Real, CoW AMM's MEV-capturing, LP-protective design. Real, programmable orders allowing more flexible trading strategies.
Pros
- Genuinely distinctive CoW matching and MEV-capturing AMM design
- Programmable orders for more flexible trading strategies
Cons
- None significant found in our research
Access only through swap.cow.fi directly, and verify the domain carefully given the platform's 2025 DNS-hijacking history.
Given the confirmed domain-hijacking incident, bookmark the official URL directly, watch for official security advisories before connecting your wallet, and consider a hardware wallet for meaningful balances.
Genuinely brilliant core engineering, undercut twice by the parts around the edges.
CoW Protocol's actual design deserves real, substantial credit: Coincidence of Wants matching and the CoW AMM are genuinely sophisticated, thoughtfully engineered solutions to real problems (MEV extraction and LP losses to arbitrage) that most of the category still hasn't solved as elegantly. In both of its confirmed incidents, that core design held: the 2023 exploit only reached a fee buffer explicitly designed to absorb solver-level risk, and the 2025 domain hijacking never touched the actual settlement contracts at all. We think that resilience is genuinely meaningful and said so directly in the score. What we can't do is pretend two real incidents didn't happen just because the smart contracts survived both; a domain hijacking that costs real users $1.2 million is a real security failure, even if it's a different kind than a contract exploit. Combined with a genuinely narrow chain footprint relative to this series' other aggregators, the fair picture is a platform with some of the best underlying engineering we've reviewed, paired with real, disclosed operational gaps around it.
The scorecard above is deliberately general. Whether CoW Swap is right for you depends heavily on which of these you already are.
The Ethereum or Gnosis Chain trader making larger, MEV-sensitive trades who wants genuinely structural price protection
This is exactly where Coincidence of Wants matching and batch auctions deliver their real, substantive value.
The liquidity provider interested in CoW AMM's MEV-capturing design specifically
The real, disclosed surplus redistribution to LPs is a genuinely distinctive benefit worth understanding.
The user who bookmarks the official domain directly given the platform's confirmed 2025 DNS-hijacking history
Given what actually happened, this specific habit genuinely matters more here than at platforms without that history.
Traders needing multi-chain reach beyond Ethereum and Gnosis, or making small, simple swaps where speed matters most
1inch or Rango, both reviewed earlier in this series, offer meaningfully broader chain coverage for this specific need.
The scorecard covers the headline judgment calls. These four tables cover the specifics we didn't want to bury in prose; both incidents timelined precisely, Coincidence of Wants and CoW AMM mechanics explained, a source-quality issue we caught and corrected, and our first seven-way aggregator comparison.
Two incidents, precisely separated from the core protocol
| February 2023 | April 2025 | |
|---|---|---|
| What happened | A malicious solver exploited an intermediary contract bug | Attackers socially engineered the domain registrar to hijack CoW Swap's DNS |
| Amount | ~$166,000-$181,600 | ~$1.2 million |
| What was affected | Protocol fee buffer only; not user funds | Users redirected to a phishing site |
| Core protocol status | Confirmed unbreached | Confirmed unbreached (contracts, APIs, solvers all secure) |
| Resolution | Solver's bond covered the damages | Domain control restored; not clear if users were separately compensated |
We're presenting these two incidents side by side specifically because both share the same reassuring pattern: the core, on-chain settlement logic held in both cases, even though real money was lost each time through different points of failure.
Coincidence of Wants and CoW AMM, explained
| Detail | |
|---|---|
| Coincidence of Wants | Two opposing orders (e.g. ETH-for-DAI and DAI-for-ETH) settle directly against each other, bypassing external liquidity entirely |
| Batch auctions | Remaining, unmatched orders are grouped and solvers compete to find the best combined execution |
| CoW AMM | A separate AMM design that captures rebalancing/arbitrage surplus and redistributes it to LPs instead of MEV bots |
This is a genuinely different mental model than a typical aggregator; you're not just getting the best quote across pools, you're potentially avoiding the pool entirely if another trader wants the opposite trade at the same time.
A source-quality issue we caught and corrected
| Detail | |
|---|---|
| The issue | One source's paragraph opens describing CoW Swap, then describes features (1inch Shield, Fusion architecture, 1inch Wallet) that clearly belong to 1inch, not CoW Swap |
| What we did | Did not attribute "nine audits" or 1inch-specific features to CoW Swap; excluded this source's specific claims from our research |
We're flagging this directly because it's a genuine, instructive example of why we verify claims across multiple sources rather than trusting any single one, even when it initially reads as directly on-topic.
Seven aggregators, side by side
| 1inch | Jupiter | Rubic | KyberSwap | Rango | Velora | CoW Swap | |
|---|---|---|---|---|---|---|---|
| Confirmed incidents | One | One | Two | One | None found | None found | Two |
| Chain scope | 13+ | Solana-only | Disputed, wide | 17-25 | 50-74+ (disputed) | Cross-chain focus | 2 (Ethereum + Gnosis) |
| Distinctive mechanism | Fusion intents | DeFi superapp | Embeddable SDK | Dual aggregator+AMM | Diamond Pattern | Delta v2.5 intents | Coincidence of Wants |
CoW Swap has the narrowest chain footprint of any aggregator in this comparison, but arguably the most structurally distinctive execution model, a real trade-off between breadth and depth.
We don't just want to hand you our number; we want to show you how it sits next to what other review desks and comparison sites have published. We've excluded one source with a demonstrated content-mixing error (attributing 1inch's features to CoW Swap), one source with embedded promotional content for an unrelated product, and all incident-specific news reports covering the two hacks, since those are postmortems rather than general reviews.
Our score lands meaningfully below the aggregated industry average; most general reviews emphasize CoW Protocol's genuine technical sophistication and MEV protection heavily, while giving comparatively less weight to its two confirmed incidents and narrow chain scope than our methodology does.
| Source | Score | Type |
|---|
Scores compiled by our editorial team from publicly available reviews as of August 2026. "Editorial estimate" means the outlet didn't publish a single numeric score, so we converted their published verdict and sentiment into a comparable 100-point figure. Verify current figures directly with each source before citing them elsewhere.
Yes, twice: a February 2023 solver exploit drained ~$166,000-$181,600 in protocol fees (not user funds), and an April 2025 domain-hijacking attack redirected users to a phishing site for ~$1.2 million in losses. In both cases, the core on-chain protocol itself was confirmed unbreached.
When two traders' orders can settle directly against each other (e.g., one wants to sell ETH for DAI while another wants to sell DAI for ETH), CoW Protocol matches them directly, avoiding external liquidity pools entirely and eliminating AMM fees, price impact, and slippage on the matched portion.
A separate AMM design that captures the rebalancing and arbitrage surplus normally extracted by MEV bots and redistributes it to liquidity providers instead, resulting in meaningfully higher LP returns than equivalent Uniswap V2-style pools in early results.
Ethereum and Gnosis Chain specifically, a genuinely narrower footprint than most other aggregators in this series, which often span 13 or more chains.
You only pay a fee if your trade executes successfully; gas costs are bundled into that fee. Order cancellations are free, and you don't need ETH upfront to sign and submit an order.
Not necessarily. Multiple independent sources note that for small, simple swaps in already-deep pairs, a direct AMM swap may be faster and simpler; CoW Swap's real advantages concentrate more in larger or MEV-sensitive trades.
Independent, competing entities that find the best execution path for batched orders. Solvers post a bond that covers damages if they act maliciously or introduce a bug, a safeguard that functioned as intended during the 2023 incident.
No. Multiple sources confirm the on-chain contracts, backend APIs, and solver networks all remained secure; the attack occurred entirely at the domain-registrar level through social engineering, redirecting users to a fake site rather than exploiting any code.
More Reviews
1inch – DEX Aggregator Review
Score: 75/100. Category-defining aggregator; a Mar 2025 resolver incident (~$5M) didn't touch user funds, but did touch trust.
Read More
Jupiter – DEX Aggregator Review
Score: 78/100. Highest score in this whole project: ~95% Solana share, with the one 2024 phishing incident purely user-side.
Read More
Rubic – DEX Aggregator Review
Score: 51/100. Lowest in this sub-series: two 2022 hacks and unresolved, wide gaps in its own claimed audit scope.
Read More
KyberSwap – DEX Aggregator Review
Score: 63.5/100. Dual aggregator+AMM; the Nov 2023 Elastic exploit left the aggregator itself confirmed unaffected.
Read More