2023 · United States · K - Financial and insurance activities · Fund theft via smart contract exploit (read-only reentrancy)
Analysed 2026-11-06 by Hermes
Open this case in the tool →Fund theft via smart contract exploit (read-only reentrancy)
Expertise to Execution
Evidence coverage: 1/6 stages, weight 8/60
Decisive stage: entry (clearest evidence) · Alternative: Strategy to Execution
(1) As a lending protocol holding user collateral on zkSync Era, EraLend's strategic expectation was a secure smart-contract architecture verified by professional audit - the protocol had undergone a pre-incident PeckShield audit presented as assurance of its robustness. (2) Operationally, an unattributed attacker on 25 July 2023 used flash loans and a read-only reentrancy against EraLend's SyncSwap-based price oracle - burning LP tokens so the callback ran before SyncSwap updated its reserves - to inflate collateral valuation and repeatedly borrow against it, draining about $2.76 million from the USDC pool; detection, containment (same-day contract pause) and disclosure were prompt, so the breach moment sits entirely on the entry gap. (3) This is a knowing-doing gap within the material column: read-only reentrancy was a documented vulnerability class, SyncSwap's burn-then-callback behaviour had already been reported, and standard mitigations (check-effects-interactions, not pricing collateral from unprotected external reserves) were available to smart-contract specialists; the protection nevertheless was never realised in EraLend's own oracle integration and the pre-incident audit missed it. The expectation originated with technical specialists (material), and the chain broke at execution - no source shows leadership deferring, under-resourcing or accepting the risk. The classification rests on the entry stage (clearest evidence); the entry vector itself is fully documented, so it is load-bearing here, unlike cases where the vector is undisclosed. (4) RM13 (Expertise to Execution lens): EraLend's team consumed a third-party DEX's LP token value as a lending price oracle without verifying or guarding its callback behaviour - trust in the SyncSwap dependency was not matched by assessment, and SyncSwap declared the callback intentional and disclaimed responsibility while EraLend only later delisted SyncSwap LP as collateral (causal role: Enabling). RM12 (Expertise to Execution lens): the pre-incident PeckShield audit underwrote the protocol's robustness claims yet the very oracle-dependency flaw later exploited passed it unnoticed, so the assurance materially misrepresented control effectiveness and left no reliable evidence by which to challenge it (causal role: Obscuring).
RM13 — Third-party / supply-chain trust boundaryRM12 — Assurance / oversight credibility gapEuRepoC row; country United States taken from eurepoc_extra.receiver_country - EraLend is a decentralised protocol with no clearly documented public headquarters, so the country rests on the dataset, not independent reporting. Initial estimates put the loss at $3.4M (BlockSec); EraLend confirmed ~$2.76M from the USDC pool. The row description has a typo (Era Lenda) but otherwise matches the victim. EuRepoC intelligence_impact records no data breach - correct: only funds were taken. mitre_initial_access Exploit Public-Facing Application used as context only. The public letter offering the exploiter a 10% bounty and to cease the investigation in exchange for 90% of funds (deadline 27 July, unanswered) is recorded as response conduct; not coded as a gap because no source documents a governance failing in that decision. No related rows in the case file. Link check: crypto-economy.com was unreachable at check time (timeout); the page was opened and read during research. Link check: coincu.com was unreachable at check time (timeout); the page was opened and read during research.