deniz.in

Markets

Weather

Loading weather

· via dev.to (home feed)

Kava Lend lost $3.76M to an accounting flaw in a dormant BUSD market

An attacker exploited a lending-accounting flaw on Kava Lend to borrow $3.76 million in KAVA tokens, after a dormant BUSD market's supply index jumped roughly 243,000-fold.

Kava Lend lost $3.76M to an accounting flaw in a dormant BUSD market

What happened

On November 14, 2026, Kava Lend — a decentralized lending application also known as HARD Lend, running on the Kava blockchain — was drained of roughly $3.76 million in a single exploit. According to blockchain analytics firm Bitquery, whose findings are laid out in a technical write-up on dev.to, the attacker borrowed about 55.2 million KAVA tokens, worth approximately $3.76 million at pre-attack prices, and retained around 53.9 million KAVA after settling some earlier loans.

The bulk of the draining took roughly 98 minutes. By the end of it, the pool's available KAVA balance had fallen from about 55.3 million tokens to 519 KAVA, and the market price of KAVA reportedly declined nearly 40 percent in the aftermath.

A dormant market, an out-of-control index

The exploit did not require breaking the Kava blockchain, compromising private keys, or minting new coins. It targeted the protocol's internal bookkeeping, specifically in a BUSD lending market that had gone quiet. BUSD, the dollar-pegged stablecoin issued by Paxos, stopped new issuance in 2023, and the associated market on Kava Lend became largely inactive.

Bitquery's investigation traced the problem to the calculation of that market's supply interest index — the accounting construct that tracks how deposited balances grow as interest accrues. When the effective supply feeding the formula shrinks to a very small number, an insufficiently guarded calculation can swing the index by extreme amounts. In this case, the affected market's supply index reportedly rose about 243,000-fold. The protocol's records then credited deposits with collateral far exceeding the assets actually present. The tokens themselves had not become more valuable; the accounting was simply wrong.

Why the usual safeguards failed

Lending protocols defend their pools with borrowing limits, collateral requirements and liquidation mechanisms. During this incident all three kept operating — but they were reading from an accounting state that no longer matched reality. Each individual transaction satisfied the application's programmed checks, yet the sequence as a whole pulled KAVA out of the pool without leaving collateral of equivalent value behind.

That is the core lesson the write-up draws: protective mechanisms that share the same flawed data source fail together. A borrowing cap is only meaningful if the collateral balances it references are trustworthy.

What Kava confirmed

On November 16, Kava published an incident report confirming that approximately 53.9 million KAVA had been taken from HARD Lend. The team stressed that the damage was confined to the lending application: the underlying blockchain was not compromised, and no additional KAVA tokens were created. That rules out an "infinite mint" style attack, in which an attacker generates new tokens outright; here, existing tokens were extracted from the pool under false accounting pretenses.

It also highlights a structural point about DeFi risk. The security of a blockchain and the security of applications built on top of it are separate concerns, and a sound chain does not prevent an application-layer financial bug.

Why it matters

For developers, the incident is a case study in how edge cases become attack surfaces. A market with almost no effective supply is easy to overlook in testing, yet that near-zero state is exactly what made the index calculation explode. Financial code needs to be exercised under extreme parameter ranges — tiny supplies, abandoned markets — not just typical operating conditions.

For protocol designers, it underlines the danger of leaving dormant markets live. The BUSD market on Kava Lend had little practical reason to remain an active part of the protocol after BUSD issuance ended, but it remained exploitable.

And for anyone auditing DeFi systems, the takeaway is that per-transaction correctness is not enough. Every operation in this exploit passed its individual checks; the failure was systemic, rooted in the economic assumptions those checks collectively depended on. Audits and invariant testing should ask not only whether each operation can be gamed, but whether a sequence of valid operations can leave the protocol insolvent.

  • #defi
  • #blockchain
  • #security
  • #kava
  • #lending

Related posts