The short answerHarvest was one of the first auto-compounders and still runs vaults across several EVM chains. Its October 2020 flash-loan exploit cost users roughly $24m, and it charges a 30% performance fee — the highest in this category — which is a difficult combination to recommend.
Best forLong-running auto-compounding on EVM chains
Strategy disclosureweight 25%
7.0avg 8.0
Contract securityweight 25%
5.0avg 7.5
Net yieldweight 20%
6.0avg 7.4
Risk controlsweight 15%
6.0avg 7.5
Feesweight 15%
5.0avg 7.4
this service category average across 15Score = Σ mark × weight. Real drawbacks; only for readers who need this one feature. Methodology.
| Criterion | Mark | What we found | Source | Checked |
|---|
| Strategy disclosure | 7.0 | not sourced yet | not sourced yet | — |
| Contract security | 5.0 | not sourced yet | not sourced yet | — |
| Net yield | 6.0 | not sourced yet | not sourced yet | — |
| Risk controls | 6.0 | not sourced yet | not sourced yet | — |
| Fees | 5.0 | not sourced yet | not sourced yet | — |
What we liked, and what fell short
Liked- Still operating after five years and multiple market cycles
- Broad vault selection across EVM chains
- Partial reimbursement was made after the 2020 exploit
Fell short- The October 2020 flash-loan exploit cost users roughly $24m
- A 30% performance fee is the highest in this category
Why the marks landed where they did
Harvest's October 2020 exploit is one of the defining incidents in DeFi's history. An attacker used flash loans to manipulate the price of stablecoins inside a Curve pool that Harvest's strategy relied on, repeatedly draining value from the vaults — roughly $24m in total, of which a portion was returned.
What the incident taught the sector
It demonstrated that a yield strategy inherits every price assumption of every protocol it touches, and that flash loans make manipulating those assumptions cheap. Most modern strategies now use time-weighted or multi-source pricing specifically because of attacks like this one.
Where it stands now
Harvest continues to operate vaults across several EVM chains with no repeat incident. Its contracts have been revised and audited. But the protocol lost its position in the market and has not regained it, and the 30% performance fee is the highest in a category where the leader charges 4.5%.
The value question
A 30% fee on yield needs to be justified by outperformance that Harvest does not demonstrate. For the same underlying strategies, Beefy takes roughly a sixth as much.
Who should use it
Few people. Users with a specific vault only available here, at small size. For general auto-compounding, cheaper protocols with cleaner records are available on every chain Harvest operates on.
FAQ
What happened to Harvest Finance in 2020?+
A flash-loan attack manipulated stablecoin prices inside a Curve pool its strategy depended on, draining roughly $24m from the vaults. A portion was returned by the attacker.
What does Harvest charge?+
A 30% performance fee, the highest in this category, against roughly 4.5% at Beefy.
Is Harvest safe now?+
Contracts have been revised and audited with no repeat incident, but the combination of that history and the fee level makes it hard to recommend.
What is a better alternative?+
Beefy for cheap auto-compounding on the same chains, or Yearn for documented strategy transparency.
