← Breach analysis by the Dual Vulnerabilities framework

Nexera token theft via admin credential compromise (Unattributed, 2024)

2024 · Netherlands · K - Financial and insurance activities · Crypto-asset theft via compromised admin credentials and smart-contract takeover

Analysed 2026-11-06 by Hermes

Open this case in the tool →

What happened

Crypto-asset theft via compromised admin credentials and smart-contract takeover

Dual vulnerability type

Expertise to Execution

Evidence coverage: 3/6 stages, weight 25/60

Decisive stage: entry (clearest evidence) · Alternative: Strategy to Execution

Why this classification

(1) Nexera is a DeFi/blockchain-infrastructure protocol whose Fundrs staking and vesting contracts, across several chains, held the keys to its entire NXRA token supply; the specialist expectation for such a protocol is that smart-contract admin credentials are kept in hardened custody (hardware wallets, multi-sig) away from ordinary office endpoints, and Nexera itself documented that the Fundrs contracts had been rigorously audited. (2) Operationally, malicious code (verified by the victim as BeaverTail malware) executed on a machine at 02:10 UTC on 7 August 2024 and sat undetected for about three hours, yielding the credentials that manage the Fundrs contracts; the attacker used them to transfer contract ownership on Ethereum and Avalanche and to transfer 47.24M NXRA, then blocked Nexera from re-assigning ownership, forcing a pause of the token across four chains and redeployment of the contracts. (3) This is a knowing-doing gap within the material column: the remedies were known technical practice, not strategic commitments - the victim's own post-mortem names hardware-wallet/multi-sig custody, endpoint and key-handling hygiene, and brought-forward staff training as what will now be put in place - and no source shows leadership deferring, under-resourcing or accepting the gap, so the chain broke at execution, in operations and architecture, not at the decision level. The classification rests on the entry stage (clearest evidence: the victim's primary post-mortem documents both the origin and the break); containment and detection point to the same type; recovery and response show positive alignment (fast pause, token burn, same-day disclosure), and treasury-level monitoring worked (alert within one minute of the drain), so they are not decisive. The case file's framing of a proxy-contract exploit overstates a code flaw: per the post-mortem the audited contracts were not the issue; the proxy was merely the reachable control point once the admin credentials were stolen. The entry vector (how the malware arrived) is not fully documented and is not load-bearing. (4) RM22 (Expertise to Execution lens): the credential-management weakness let the attacker act as a legitimate administrator - contract-admin credentials were held and used on an internet-connected ordinary machine without the hardware-wallet/multi-sig protection the victim itself planned to introduce afterwards, a custody control specialists in this sector know and Nexera had not realised (causal role: Enabling). RM28 (Expertise to Execution lens): endpoint protection around the key-management machine was absent or ineffective - publicly documented, non-zero-day malware (BeaverTail) ran for roughly three hours, harvested the admin credentials and escalated, a hardening and malware-protection step known but not executed (causal role: Enabling). RM04 (Expertise to Execution lens): a single stolen credential set concentrated control of the staking contracts on Ethereum and the vesting contracts on Avalanche; once hijacked it blocked Nexera's re-acquisition of ownership, so one endpoint compromise amplified into the full NXRA supply at risk, a multi-chain token pause and forced contract redeployment, reducing recovery options (causal role: Amplifying).

Evidence by stage

Risk markers

Sources

Notes

Sources: victim post-mortem (primary, opened 06-11-26), HackRead (secondary, opened), crypto.news (secondary, opened). The case file's Cyvers X post could not be opened directly and the Cyvers blog article returned empty; Cyvers' alert is quoted via the HackRead article, which was opened. Figures: initial reporting said ~$1.8M (32.5M NXRA + ~$555K USDT held by the attacker); the victim's post-mortem says 47.24M NXRA were stolen, the attacker sold 14.75M (~$449K), and the remaining 32.5M were removed from the attacker's wallet and burned (crypto.news); the EuRepoC sheet records $1.7M economic impact. Attribution: dataset says Not attributed; the post-mortem verified BeaverTail (and latent InvisibleFerret) malware whose deployment is consistent with state-backed (DPRK) actors per cited Unit 42 research; ZachXBT linked the wallet to earlier private-key compromise exploits (SpaceCatch, Concentric Finance, OKX DEX, Serenity Shield, Reach) - pattern evidence only, not incident-specific attribution. Earlier incident at the same victim (as AllianceBlock, 2022: 110M legacy ALBT tokens lost via the Bonq borrowing protocol, per crypto.news) used a different mechanism (token/protocol exploit, not credential compromise), so no recurring-weakness marker is coded. Description check: the case-file description attributes the theft to an exploit in the proxy contract; the victim's own post-mortem states the audited smart contracts were not the cause and the root cause was credentials stolen from a malware-infected machine - the classification follows the primary post-mortem. Country from eurepoc_extra.receiver_country (Netherlands), consistent with the post-mortem's contact with the Dutch police; the victim (formerly AllianceBlock) is Amsterdam-founded.