OAK — OnChain Attack Knowledge

Worked example · 2024-07

Puffer pufETH LRT depeg — redemption-constrained discount — Ethereum — 2024-08

Loss
structural — per LlamaRisk, pufETH's maximum depeg against ETH was −5.46% on 2024-08-05. No aggregate dollar loss figure is sourced; the cost was borne by holders who sold (or were liquidated against pufETH collateral) during the discount window. The sources reviewed report no protocol insolvency or bad debt.
OAK Techniques observed
OAK-T14.004 (Liquid Restaking Token Pricing Manipulation / Depeg — LlamaRisk attributes the discount mainly to the difficulty of redeeming pufETH for ETH: Puffer did not enable withdrawals until 2024-10-14, so holders needing ETH could only sell on secondary markets at a discount). Renzo ezETH (2024-04-24) is a separate LRT depeg of the same year; whether ether.fi weETH and Kelp rsETH also depegged in sequence is unverified here and is not asserted.
Attribution
unattributed — structural / protocol-design-level. No attacker is involved.
Corrected 2026-10-11
Depeg date moved from July to 2024-08-05 with the sourced −5.46% maximum (not 4–7%); cause restated as difficulty of redeeming pufETH (withdrawals only enabled 2024-10-14) rather than an EigenLayer withdrawal queue; the unsourced "fourth in a sequence" framing, queue-depth claims and proposed-only references were removed.
Key teaching point
A liquid restaking token that cannot be redeemed at par trades at a discount when holders want out. pufETH had no withdrawal path until October 2024, so the secondary-market price was the only exit and the discount reflected that illiquidity, not a solvency problem. Redemption availability, not just backing, is the variable to monitor for any LRT.

Summary

Puffer Finance launched pufETH as a liquid restaking token in 2024, letting holders earn restaking yield while keeping a liquid token for DeFi use. Users deposit ETH, Puffer restakes it, and pufETH is minted as a receipt for the underlying ETH plus accrued yield.

LlamaRisk's collateral risk assessment records a maximum depeg of −5.46% on 2024-08-05 and links the discount to the difficulty of redeeming pufETH for ETH (low secondary-market liquidity may also have contributed). Puffer only enabled withdrawals on 2024-10-14, after which two routes existed: an instant redemption with a 1% fee limited by PufferVault ETH liquidity, and a fee-free two-step withdrawal taking at least 14 days. LlamaRisk reports the peg has since recovered to roughly 0.1%.

This file previously described the discount as caused by an EigenLayer withdrawal-queue backlog that cleared in late July. No source supports that mechanism or timing, so it has been removed.

Timeline (UTC)

When Event OAK ref
2024-04-24 Renzo ezETH depegs (separate LRT event) T14.004 (context)
2024-08-05 pufETH reaches its maximum depeg of −5.46% against ETH (LlamaRisk) T14.004
2024-10-14 Puffer enables pufETH withdrawals (redemption path opens)

What defenders observed

  • Redemption availability drives LRT pricing. With no withdrawal path, pufETH holders could exit only by selling at the market price; a redemption route (even a fee-bearing one) gives arbitrageurs a floor.
  • Peg recovery is not documented in detail. LlamaRisk reports the peg later improved to about 0.1%; the sources reviewed do not describe the recovery path day by day.

What this example tells contributors writing future Technique pages

  • Separate redemption-constraint depegs from oracle or collateral-failure depegs. The pufETH discount was a liquidity/redeemability effect.
  • Verify cohort claims. Sequences of "same-mechanism" depegs across LRTs should be sourced per event before being cited as a class.

Public references

Discussion

The pufETH episode illustrates T14.004 as a redemption-constraint discount rather than an exploit. OAK's wider LRT-depeg cohort claims (ether.fi, Kelp) should be verified individually before being relied on.

Techniques demonstrated (1)