Skip to content
The Canadian DeFi rulebook · rules checked 2026-11-02 · what changedEducation only · not financial, tax or legal advice
Safety

How to read a smart contract audit (without being a developer)

What a smart contract audit covers and misses, how to read severity and fix status, check admin keys and timelocks, and confirm the deployed contract matches.

Blue square · intermediate9 min readFacts checked 2026-11-02
Mitey the yeti talking a plan through with an older yeti, with a magnifying glass over code
Mitey says: An audit is a snapshot of one version of the code, so check that the snapshot matches what you’re actually using.

To read a smart contract audit, first check its scope, the code version reviewed and the review dates, then read the status of each finding. Acknowledged means the team chose not to fix an issue; fixed should come with a check by the auditor. An audit reduces risk but never removes it, because upgrades and admin keys can change code.

Trail notes

  • An audit is a time-limited review of specific code by specific people; it reduces risk but never removes it.
  • Check the scope, the code version and the date first, then read what happened to each finding.
  • “Acknowledged” means the team knows about an issue and chose not to fix it; “fixed” should come with a check by the auditor.
  • Upgradeable contracts and admin keys can change the code after the audit, so look for timelocks and shared control.
  • You can confirm on a block explorer that the deployed contract is verified and matches the audited address.

What an audit is, and what it isn’t

A smart contract is a program stored on a blockchain that runs exactly as written. As ethereum.org’s introduction explains, once deployed it executes automatically, which is why mistakes in the code can be so costly. The smart contract basics guide covers how they work.

An audit is a review of that code by an outside security firm or independent researchers, usually over a few weeks. Auditors look for bugs, logic errors, ways to manipulate prices or permissions, and departures from the project’s own documentation. They write up what they found in a report.

An audit is not a guarantee, an insurance policy or an endorsement. It covers the code the auditors were given, during the time they had, against the risks they thought to test. It usually does not cover the economics of the design, the people running the project, the website you use to connect, or code added after the review. A project can say “audited” and still lose funds.

“Audited” is a claim to verify

Scam projects sometimes link to reports for different code, publish fake reports, or cite an audit of an earlier version. Find the report on the auditor’s own website or repository, not only on the project’s site.

Start with scope, version and date

Most reports open with a summary page. Before reading any findings, find three things:

  1. Scope. Which contracts or files were reviewed. If the lending logic was audited but the price oracle adapter was not, the unreviewed part is still a risk.
  2. Version. The code is usually identified by a commit hash, a long string of letters and numbers that pins an exact version of the code. A report on one commit says nothing certain about later commits.
  3. Date and duration. When the review happened and how long it lasted. A short review of a large codebase is thinner than it sounds.

Many reports also list the contract addresses the auditors checked, or the team publishes them in its documentation. Note those addresses; you will compare them with what is deployed later.

Risk mapLikelihood for an ordinary user →Severity →ScamsBridgeSmart contractLiquidationDe-pegOraclePlatform failureLost keysRegulatorywatch closely
DeFi risk map: likelihood and severity (editorial). An editorial teaching map, not a measurement. Positions reflect how often each risk tends to affect ordinary users and how badly when it does.
  • Scams and phishing: common and often total loss.
  • Bridge failures and smart-contract exploits: less frequent for any one user, but severe.
  • Liquidations: common for borrowers in fast markets; partial loss plus penalties.
  • Stablecoin de-pegs and oracle problems: occasional; severity depends on design and exposure.
  • Platform failure and lost keys: can be total; mostly preventable with checks and backups.
  • Regulatory change: frequent but usually lower direct loss.

Reading findings: severity and status

Findings are usually graded by severity. Labels vary between firms, but a common scale looks like this:

Severity names and definitions differ between auditors; read each report’s own definitions page.
SeverityWhat it usually means
CriticalFunds can be stolen or permanently locked, often with little effort.
HighSerious loss is possible under realistic conditions.
MediumLoss or disruption is possible under specific conditions.
LowMinor issue, limited impact, or best-practice gap.
InformationalCode quality, documentation or gas-efficiency notes.

The status of each finding matters as much as the severity. Fixed or resolved means the team changed the code; a good report says the auditor reviewed the fix. Acknowledged means the team accepted the finding but chose not to change the code, sometimes for good reasons, sometimes not. Partially fixed or mitigated means the risk was reduced, not removed.

Example: reading one report (hypothetical)
A report lists 1 high, 3 medium and 6 low findings.
The high and two mediums are marked fixed and re-checked by the auditor.
One medium, about a single admin being able to change collateral settings instantly, is acknowledged.
Your takeaway: the code bugs were addressed, but a governance risk remains by design. That is something to weigh, not ignore.

Multiple audits, bug bounties and formal verification

More than one audit by different firms, especially when each covers the current version, gives you more independent eyes on the code. Repeated audits of the same old version add less than they appear to.

A bug bounty is a standing offer to pay researchers who report vulnerabilities privately. It encourages ongoing scrutiny after launch, which a one-time audit cannot provide. Check whether the bounty covers the contracts you would use and whether it is still active.

Formal verification is a mathematical method of proving that code behaves according to a written specification, for every possible input rather than only the cases someone tested. It is powerful for well-defined properties, such as “total deposits always equal the sum of balances”, but it only proves what the specification says. If the specification misses a risk, the proof will too.

Upgradeability, admin keys and timelocks

Many protocols are upgradeable. A proxy contract holds your funds and points to a separate implementation contract containing the logic. Whoever controls the upgrade function can point the proxy at new code, which means the audited code can be replaced.

Look for answers to these questions in the documentation or the report:

  • Who holds admin or owner permissions: one key, a multi-signature group, a governance vote?
  • What can those permissions do: pause withdrawals, change fees, change collateral settings, upgrade the code?
  • Is there a timelock, a built-in delay between announcing a change and applying it, so users have time to react and withdraw?
  • Can anyone mint tokens or move user funds directly?

A single key with instant upgrade power is a serious concentration of trust, regardless of how clean the audit was. The smart contract risk page explains how these risks combine, and rug pulls and scams shows how admin powers have been abused.

Checking the deployed contract yourself

You do not need to read code to do a basic match check on a block explorer, a public website that shows every address and transaction on a network:

  1. Get the contract address from the project’s official documentation, not from a social media post or search ad.
  2. Search that address on a block explorer for the correct network.
  3. Look for a verified source code badge or tab. Verified means the published code compiles to the bytecode actually deployed at that address.
  4. Compare the address with the one listed in the audit report or the team’s audited deployments page.
  5. If the explorer labels it as a proxy, note the current implementation address and check whether that implementation was audited.
  6. Check recent admin transactions, such as upgrades or ownership changes, and their dates relative to the audit.

If the code is unverified, the address does not appear in any report, or the implementation changed after the audit, treat the audit as not covering what you are using. The protocol risk scorecard helps you record these checks in one place. Before approving any contract from your wallet, review wallet approvals and signing.

Red flags at a glance

No audit, or an audit you cannot find on the auditor’s own site. Critical or high findings left acknowledged. An audit of an old version. Unverified contracts. A single admin key with no timelock. A team that will not say who controls upgrades.

What this means for you

  • Find the audit report on the auditor’s own website or repository, and read the status of every critical and high finding.
  • Check who holds admin keys and whether a timelock delays upgrades, giving users time to withdraw before changes take effect.
  • Confirm on a block explorer that the contract is verified and its address matches the report, and for proxies that the current implementation was audited.
  • Look for an active bug bounty covering the contracts you would use; see smart contract risk for other ways code fails.

Your checklist

Questions people ask

Does a smart contract audit mean a protocol is safe?

No. An audit is a time-limited review of specific code and lowers risk, but it does not guarantee safety. Bugs can be missed, code can change after the audit, and audits usually do not cover economic design, governance or the people running the project.

What does “acknowledged” mean in an audit report?

Acknowledged means the project team accepted that the issue exists but chose not to change the code. It may be a deliberate design choice or an unresolved risk, so read the team’s explanation and weigh it, especially for medium or higher findings.

How do I check that a deployed contract matches the audit?

Look up the contract address from the project’s official documentation on a block explorer, confirm the source code is verified, and compare the address with the one in the audit report. For upgradeable proxies, also check that the current implementation contract was audited.

Why do timelocks matter?

A timelock forces a delay between announcing a change and applying it. That gives users time to see an upgrade or settings change and withdraw before it takes effect, which limits the damage a compromised or careless admin key can do.

Sources and dates

  • Introduction to smart contracts — ethereum.orgreference
    No date shown on page · checked 2026-10-31. What a smart contract is and how it executes.
  • Ethereum security and scam prevention — ethereum.orgreference
    No date shown on page · checked 2026-10-31. Wallet security habits and common scams.
  • Decentralized finance (DeFi) — ethereum.orgreference
    No date shown on page · checked 2026-10-31. Overview of DeFi building blocks on Ethereum.
  • Oracles — ethereum.orgreference
    No date shown on page · checked 2026-10-31. How off-chain data reaches smart contracts, and oracle risks.

Education only. Nothing on this page is financial, investment, tax or legal advice, or a recommendation to buy, sell, stake, lend or borrow any crypto-asset. Crypto-assets can lose some or all of their value and are not covered by CDIC. Canadian rules change: check the dated sources above and talk to a qualified professional about your situation. Full disclaimer.

Rule-change alerts

Ask to be told when a rule on the tracker changes status. The MITEY network team sends these, and your enquiry goes to the MITEY intake inbox. For instant updates use the RSS feed or the JSON file.