What a Good Smart Contract Audit Report Contains (and How to Spot a Weak One)
Security Wiz team3 min read
An audit report is the product you're buying. It's what your engineers act on, what your community reads, and what an investor may ask to see. A good one reads like a thorough home inspection: what was checked, what was wrong, and the evidence. So it's worth knowing what a good one looks like, and what to watch for.
What a good report contains
A summary anyone can read. What was reviewed, what was found, and where the project stands, written for someone who isn't an engineer.
Exactly what was reviewed. Which files, at which exact version of the code. Without this, the report doesn't say what it covers.
A threat model. A short description of what could go wrong: the money or assets that could be stolen or frozen, the people involved (users, administrators, automated bots, attackers who can borrow huge sums for a single transaction), and what the project is trusting each of them to do. It's how you can tell the auditor understood your product, and not products in general.
Testing against a known standard. For us, that's a table covering all ten categories of the OWASP Smart Contract Top 10, showing how each was tested, including the ones where nothing was found. "Tested, nothing found" is useful information.
Findings, each with:
- a severity, and the inputs behind the score so you can check it (for problems in the contract code we use OVSS)
- the OWASP category it belongs to and the standard ID for that type of weakness
- a description of the problem and what it could cost you, in plain terms
- proof: a test you can run yourself that shows the problem happening or, where a test isn't practical, the exact route through the code that causes it
- a recommended fix
A retest section, showing where every finding stands after your fixes: fixed, acknowledged (you've accepted the risk), or disputed.
A grade, broken down by category, so readers can see where the project is strong and where it isn't.
Warning signs of a weak report
- It looks like a printout from a scanning tool. Pages of automated results with generic descriptions and no discussion of how your product actually works. Scanner output is a list of leads, not findings.
- Findings without proof. "This could be exploited," with no test, no walkthrough, and no scenario. If a problem can't be reproduced, you can't tell a real one from a false alarm.
- Severity with no reasoning. A label, and no explanation of why it's High and not Medium. A good report shows its reasoning, and the inputs behind the score let you check it.
- Nothing about what wasn't found. No coverage table, no list of what was tested. You can't tell whether a short list of findings means clean code or a shallow review.
- No version and no list of what was reviewed. You can't tell what the report is about.
- No retest. A report that ends after the first round never confirms that the fixes worked, and fixes are where new bugs often creep in.
- A report you can't publish. If the firm won't let you share it, or it's written so nobody outside your team can follow it, it won't do the job of reassuring your community.
Why publishable matters
Your users and investors can't read your code, but they can read a report. A report written to be shared, with the sensitive parts held back until they're fixed, turns an audit from an expense into something that builds trust. That's the standard we write to: clear enough to post next to a launch announcement, and honest enough to stand up to scrutiny.
If you'd like to see what a report for your project would look like, request an audit, or read how our audits run.
Shipping something onchain?
Send us your repo and we’ll return a quote within 24 hours.