A smart contract audit is a review of a protocol’s code by an outside security firm, looking for bugs that could let someone steal or freeze funds. The auditor reads the contracts, tests them, lists every issue by severity and publishes a report. An audit lowers the odds of a code exploit. It does not make a protocol safe, and most of the biggest DeFi hacks of the past two years hit audited code.
That gap between “audited” and “safe” costs users hundreds of millions a year. So this guide covers what an audit checks, what it leaves out, how to read an audit report in five minutes and how I weigh one before putting farming capital into a protocol. It follows our white hat vs black hat guide, which covers who finds the bugs the auditors missed.
What a smart contract audit actually does
An audit is a paid engagement, usually two to six weeks, between a protocol and a security firm. Trail of Bits, OpenZeppelin, Spearbit, Halborn and CertiK are the names you will see most. Prices run from a few thousand dollars for a simple token to several hundred thousand for a lending protocol or bridge.
The work has three layers. First, manual review, where engineers read every line looking for known bug classes: reentrancy, integer overflow, broken access control, faulty math. Second, automated tools that scan for the same patterns at scale. Third, testing, where auditors write attack scenarios and run them against the code.
The result is a report. Each finding gets a severity rating, from critical down to informational, and a status, either fixed, acknowledged or disputed. A clean audit means every critical and high finding was resolved before launch. It does not mean nothing was found.
Some protocols add formal verification, which mathematically proves specific properties of the code. Others run audit contests on platforms like Code4rena or Sherlock, where hundreds of researchers compete for a prize pool. Both help. Neither is standard.
Why audited protocols keep getting drained
The uncomfortable list from the last two years is long. Balancer was one of the most audited protocols in DeFi and lost $128 million in November 2025 to a rounding error. Cetus passed audits and lost $223 million to an overflow bug in May 2025. Flooring Protocol survived multiple reviews and still got drained through gas-optimised code that hid the flaw from every reviewer, as our Flooring Protocol exploit post shows.
Five reasons explain the gap.
Scope. An audit covers the contracts listed in the engagement, at the commit hash on that date. Anything added, upgraded or redeployed afterwards is unaudited until someone pays for another review. Many protocols ship a dozen upgrades on one audit badge.
Time. Auditors get weeks. Attackers get years, and they only need one path the auditors did not walk.
Economics. Audits check code, not incentives. Oracle manipulation, flash loan attacks and governance takeovers often use the code exactly as designed. Nothing is “broken”, yet the funds leave.
Infrastructure. The two biggest DeFi hacks of 2026 had nothing to do with contract bugs. Drift lost $285 million through stolen admin keys, and KelpDAO lost $292 million through poisoned RPC nodes feeding its bridge false data. Our KelpDAO exploit explainer walks through how a bridge that passed review still got lied to. No audit report covers a developer’s laptop or a node provider.
Tooling. Attackers now point AI models at old, audited code and let them grind through edge cases no human reviewer caught. The Zcash Orchard flaw was surfaced this way, and our post on AI crypto exploit risk explains why a 2023 audit says less every year.
The audit badge problem
Walk through any launchpad and you will see “Audited by CertiK” on half the tokens. Plenty of them rugged anyway. The badge is real. It just does not mean what people think.
A token audit typically checks that the contract does what it says: no hidden mint function, no blacklist, no fee switch the owner can flip. It says nothing about whether the team will dump, whether liquidity is locked or whether the project has a product. Rug pulls are a business decision, not a bug, and our guide on crypto rug pulls covers the signals that matter there.
Worse, some “audits” are a few hundred dollars for an automated scan with a PDF attached. An audit from a firm nobody has heard of, finished in two days, with zero findings, is marketing. Real reviews find things. A report with no medium or high findings on a complex protocol should make you more suspicious, not less.
How to read an audit report in five minutes
You do not need to understand the code. You need to check five things.
Step 1. Find the date and the commit hash. Compare the date to the protocol’s last major upgrade. If the upgrade came after the audit, the live code is unaudited.
Step 2. Check the scope. The report lists which contracts were reviewed. If the vault you are depositing into is not on the list, it was not audited.
Step 3. Count the findings by severity. A complex protocol with one or two critical findings, all marked fixed, is normal. A protocol with critical findings marked “acknowledged” has shipped with known holes.
Step 4. Look for “acknowledged” and “disputed” statuses. Those are the issues the team chose not to fix. Read why.
Step 5. Check who paid. An audit commissioned by the protocol is standard. One commissioned by a launchpad as part of a listing package is often shallow.
Ten minutes of this tells you more than any badge on a website.
How I weigh audits when farming
Airdrop farming puts capital into new protocols by design, so I cannot avoid audit risk. I can size it.
An audit from a top-tier firm, dated after the current deployment, with critical findings resolved, earns a protocol a normal farming position. No audit, or an audit from a name I cannot find anything about, means dust-level exposure only. A protocol that has already survived a public bug bounty programme and a few months of mainnet gets more.
Even then, the audit only covers the code. Before I deposit, I also read the docs for admin key setup, check whether a multisig or timelock guards upgrades, and look at how the protocol handled its last incident. Our guide on whether airdrop farming is still safe covers that wider checklist, and the which DEX is safe comparison applies the same questions to the big perp venues. Every exploit that beat an audit this year is logged in our DeFi hacks 2026 guide, and each term above is in the crypto security glossary.
The rule underneath all of it is older than audits. Never keep more in one protocol than you can lose without changing your plans.
Keep This Content Free
Reading audit reports is nobody’s idea of fun, which is why we do it before writing farming guides. Trading capital, on the other hand, sits on exchanges that have passed a different test. OKX and Bybit both publish proof of reserves, and Bybit paid everyone back after a $1.5 billion hack. Signing up through our links gets you their welcome bonus and keeps guides like this one free.
Final Words
A smart contract audit is a snapshot of one version of the code, reviewed by a few people for a few weeks. It catches a lot. It cannot catch upgrades shipped afterwards, incentives that work as designed, stolen keys or poisoned infrastructure, and the list of audited protocols that got drained anyway keeps growing.
Read the report, check the date and scope, and treat “audited” as the start of your research rather than the end of it. Our security tag tracks every exploit as it lands, audited or not. Next in this series, supply chain attacks, the method behind Bybit, DMM Bitcoin and the Ledger reseller scandal, where the target’s own code was never the problem.
FAQ
What is a smart contract audit?
A review of a protocol’s code by an outside security firm to find bugs that could let someone steal or lock funds. The auditor reads and tests the contracts, rates each issue by severity and publishes a report showing what was fixed.
Does an audit mean a protocol is safe?
No. It lowers the chance of a code exploit in the reviewed contracts at that date. Upgrades after the audit, economic attacks, stolen admin keys and compromised infrastructure are all outside its scope.
Which audited protocols have been hacked?
Balancer ($128 million, 2025), Cetus ($223 million, 2025), Flooring Protocol (2026) and KelpDAO ($292 million, 2026) all had audits. Most large DeFi exploits since 2022 hit audited code.
How much does a smart contract audit cost?
From a few thousand dollars for a simple token to several hundred thousand for a complex protocol. Price tracks the size of the codebase and the reputation of the firm.
What is the difference between an audit and a bug bounty?
An audit is a paid review before launch by one firm. A bug bounty is an open standing offer that pays anyone who reports a flaw after launch. Strong protocols use both.
How do I check if a protocol is audited?
Look for a security or audits page in the protocol’s docs, then open the report itself. Check the date, the listed contracts and the findings. A badge on the homepage without a readable report counts for nothing.
Credit: Source link

















