On February 21, 2025, attackers drained roughly $1.5 billion from Bybit in the largest cryptocurrency theft ever recorded — not by cracking the exchange's smart contracts or brute-forcing its servers, but by tampering with a third-party interface that Bybit's own signers trusted on sight. That single event crystallized a problem security teams had been warning about for years: crypto exchange software supply chain risk is now the dominant attack surface in digital-asset infrastructure. Exchanges and DeFi platforms are stitched together from hundreds of external dependencies — wallet SDKs, price oracles, npm packages, CI/CD pipelines, multisig front-ends, and vendor APIs — and attackers have learned that compromising one weak link upstream is cheaper and quieter than attacking a hardened exchange head-on. This piece breaks down where that risk concentrates, walks through the incidents that prove it, and outlines what real defense looks like.
What is crypto exchange software supply chain risk, and why is it different from a normal hack?
Crypto exchange software supply chain risk is the exposure created when an exchange or DeFi platform inherits a compromise from something it didn't build itself — a library, a vendor tool, a CI pipeline, or a piece of open-source code pulled in during a routine npm install. It's different from a "normal hack" because the exchange's own perimeter can be flawless and it still loses funds, since the attacker never touches the exchange directly. Instead, they poison a dependency three or four hops upstream — a wallet connector, a build tool, a monitoring agent — and let the exchange's own deployment process carry the malicious code across the front door. Sonatype's 2024 State of the Software Supply Chain report counted over 512,847 malicious packages published to open-source registries that year alone, a 156% year-over-year jump, and financial services and crypto tooling are disproportionately targeted because a single successful injection can be monetized in minutes. For an exchange holding customer custody, that asymmetry — enormous payout, small blast radius to detect — is exactly why supply chain risk has overtaken smart contract bugs as the leading cause of nine-figure crypto losses.
How do blockchain wallet SDK vulnerabilities expose exchanges and their users?
Blockchain wallet SDK vulnerabilities expose exchanges because the SDK sits at the exact point where a transaction is constructed and signed, so a compromised SDK doesn't need to break cryptography — it just needs to lie about what's being signed. The clearest example is the Ledger Connect Kit incident of December 14, 2023: a former Ledger employee's npm account was phished, and attackers pushed a malicious version of the widely embedded @ledgerhq/connect-kit library that silently redirected users' wallet approvals to a drainer contract. Because Connect Kit was pulled live from a CDN by dozens of dApps — including Zapper, SushiSwap, and Revoke.cash — the compromise propagated to production sites that had no code change of their own to review. Roughly $600,000 was stolen in under 40 minutes before the package was pulled, and the incident forced the entire ecosystem to confront the fact that wallet SDKs and connector libraries are effectively shared infrastructure, not vendor-scoped code. Any exchange or DeFi front end that dynamically loads a wallet SDK from a third-party CDN, rather than pinning and vendoring a reviewed version, is exposed to the same failure mode.
Why does DeFi smart contract dependency risk extend beyond the contract's own code?
DeFi smart contract dependency risk extends beyond a protocol's own code because most contracts are composed from imported libraries, oracle feeds, and cross-chain bridge logic that the protocol team never audited from scratch. Poly Network lost $611 million in August 2021 not because its core logic was sloppy, but because a cross-chain contract call trusted an external verification step it didn't fully control. Euler Finance lost $197 million in March 2023 through a flaw in a donation function interacting with its own lending logic in an unexpected order — the kind of composability bug that only appears once a contract depends on the exact behavior of other contracts calling into it. Bridges, which are essentially the supply chain connective tissue of DeFi, accounted for over $2.5 billion in cumulative losses through 2022, per Chainalysis, precisely because they import trust assumptions from every chain and validator set they touch. Auditing a contract's own Solidity is necessary but not sufficient: a "clean" audit report says nothing about the oracle it calls, the proxy pattern it upgrades through, or the third-party math library it imported that turns out to round incorrectly at the edges.
Why do exchange platform third-party audit programs keep missing these attacks?
Exchange platform third-party audit programs keep missing these attacks because most audits are scoped to a snapshot of application code at a point in time, not to the living, constantly-updating dependency tree that ships in every subsequent release. A smart contract audit dated six months ago says nothing about the npm package that was updated last Tuesday, the CI runner image that was patched last night, or the analytics SDK that just shipped a new minor version pulling in a transitive dependency nobody reviewed. This gap is exactly how North Korea's Lazarus Group operates: Mandiant's investigation into the March 2023 3CX supply chain compromise found the attackers had trojanized a legitimate, digitally signed desktop application to reach downstream targets, several of them cryptocurrency firms, and the same group was later linked to the $37 million CoinsPaid breach in July 2023, achieved by sending a fake job-interview coding assignment laced with malware to an employee. Neither attack would have been caught by a code audit, because the audited software wasn't what was compromised — the toolchain and the humans running it were. Third-party audits remain valuable, but treating them as a substitute for continuous dependency monitoring, SBOM verification, and build-pipeline integrity checks is precisely the blind spot attackers now target by default.
What do the biggest breaches actually have in common?
The biggest breaches share a common shape: the attacker compromised something upstream of the target and let the target's own trusted process do the damage. The Bybit theft in February 2025 followed this pattern almost exactly — investigators traced the $1.5 billion loss to a manipulated user interface served through Safe{Wallet}'s infrastructure, which altered what Bybit's multisig signers believed they were approving even though the underlying smart contract logic was untouched. It echoes the 2018 event-stream incident, one of the earliest widely documented cases, in which an attacker gained maintainer access to a popular npm package specifically to reach Copay, a Bitcoin wallet application, and inject code that stole private keys from users who updated normally. Six years and dozens of incidents later — Ledger, 3CX, CoinsPaid, and now Bybit — the throughline hasn't changed: exchanges and wallets keep getting breached through code and infrastructure they trusted but never independently verified at the version actually running in production.
How Safeguard Helps
Safeguard closes this gap by treating every dependency an exchange or DeFi platform relies on — wallet SDKs, npm and PyPI packages, CI/CD runners, container images, and vendor integrations — as a continuously monitored asset rather than a one-time audit line item. Safeguard generates and verifies software bills of materials (SBOMs) across the full build pipeline, flags newly published or unusually behaving versions of libraries before they reach production, and detects the kind of silent dependency substitution that drained Ledger Connect Kit users and, at much larger scale, Bybit. For teams managing DeFi smart contract dependency risk, Safeguard tracks the provenance of imported libraries, oracle integrations, and proxy contracts so that a "clean" audit six months ago doesn't quietly go stale as the dependency tree underneath it changes. And because attackers increasingly target the humans and toolchains around the code rather than the code itself, Safeguard extends coverage to CI/CD configuration drift and signing infrastructure — the same categories of weakness exploited in the 3CX and CoinsPaid incidents. For exchanges that already run third-party security audits, Safeguard is the layer that keeps working between audits, giving security and engineering teams continuous, verifiable visibility into exactly what code — and whose code — is running in production at any given moment.