Skip to content
SECQON

How to read a pentest report: demonstrated, confirmed, likely and unconfirmed

· 7 min read · Pentest reports, How SecQon works

A pentest report arrives with forty findings, eight of them marked high. Where do you start?

Most people start with severity, and that is half right. Severity tells you how bad a finding would be if it is real. It does not tell you whether it is real. For that you need the second label on every finding: how far it was proven.

The four proof labels

Every finding in a SecQon report carries one of four labels. They describe what we actually established, not what a tool suspected.

Demonstrated

A bounded proof-of-exploit changed how your application behaved. This is the strongest label, and only a few classes of issue can earn it:

  • Reflected markup injection, rendered in a headless Chromium browser, with a screenshot.
  • SQL injection, shown by a boolean differential: the application answers differently to a true condition than to a false one. No data is read.
  • Server-side request forgery (SSRF), shown by your server making a call to a collector we run.

A demonstrated finding is not a theory. Someone with the same request can do the same thing.

Confirmed

A re-test observed the same thing again. After the first pass, every candidate finding is re-tested, and the transcript of that reproduction is attached to the finding. Confirmed is reserved for what a re-test reproduced. It is never assigned because of which tool found something.

Confirmed does not always mean exploitable. For an open port it means the port answered again. It is not a claim that the service behind it is unauthenticated. The finding's explanation says what was confirmed.

Likely

The issue was seen once and could not be reproduced, or a tool reported something serious, such as a known CVE or a CVSS score of 7.0 or more, that has not been proven. A likely finding has evidence attached and deserves attention, but treat it as a strong lead rather than a fact.

Unconfirmed

Only a tool reported it, and nothing was observed that backs it up. Unconfirmed findings are where false positives live. Read them, but do not let them displace the work above.

Reading severity and proof together

Severity runs from critical to info. Combine it with the proof label and a sensible order falls out:

Demonstrated / confirmedLikelyUnconfirmed
Critical / highFix nowInvestigate this weekCheck when you can
MediumFix this sprintInvestigateReview
Low / infoScheduleReviewUsually ignore

Scroll the table sideways to see every column.

This is a starting point, not a rule. A medium-severity issue on your login page may matter more than a high on a forgotten test host. The table answers "where do I start?" so that triage is a decision rather than an investigation.

What else to read on each finding

  • The evidence. The request that produced the finding and the relevant part of the response. If you disagree with a finding, the evidence is where you prove it.
  • The explanation. What the issue means in plain language, and what an attacker could do with it.
  • The fix. Step-by-step remediation, with configuration for common servers and frameworks where it applies.
  • The mapping. The OWASP Top 10 category, and the SOC 2 criteria it bears on, useful when the report is going to an auditor or a customer.

Read the limits section too

A good report says what it did not do. In a SecQon report, the appendix lists each stage that ran on that scan, and the report names what an automated external test cannot assess. That includes business logic, race conditions, stored XSS and chained attack paths, and the OWASP categories A04 and A09, which are not observable from outside.

That section matters for two reasons. It tells you what you still need to cover another way. It also lets whoever receives the report judge it fairly.

After you fix something

A finding is not closed because you deployed a fix. It is closed when a re-test no longer reproduces it. In SecQon a finding is an issue that outlives the scan that found it. When a later scan stops observing it, the record shows the fix with its date, and that becomes the proof you can show a customer or auditor.

Try it on your own asset

The free plan covers one verified asset with the full non-intrusive check set, and every finding carries its proof label and evidence. There is no card on file, so it is a low-risk way to see what these labels look like on your own site before anyone else asks for a report.

See what a scan finds on your own asset

Free for one verified asset, no card needed. Paid plans add the written report, more assets and deeper scans.

Start freeCompare plans

More articles