138% more reported CVEs per day in 2026 than last year.New: always-on Fixboxes. Vulnerable dependencies, fixed and verified.

Security testing

Penetration test report

The report is what you pay for in a penetration test. A good one tells executives how exposed you are in a page, tells engineers exactly how to reproduce and fix each finding, and gives auditors and customers evidence that the test happened. Below is what it should contain, a template you can hold a tester to, and what to do with the results.

Updated

What a pentest report contains

  1. Executive summary. The overall risk in plain language, the most serious findings and what they would let an attacker do. One page.
  2. Scope and rules of engagement. Exactly what was tested, from where, with which accounts, on which dates, and what was excluded.
  3. Methodology. The method followed, such as the OWASP Web Security Testing Guide, and the mix of manual, automated and AI-driven testing.
  4. Findings. One entry per finding, in the format below.
  5. Attack narrative. How findings were chained, step by step, for anything beyond a single weakness.
  6. Remediation summary. The fixes in order of priority, with owners if the tester knows them.
  7. Retest results. Added after the retest: which findings are fixed, partly fixed or still open.
  8. Appendices. Tools used, raw output and the list of components in scope.

How each finding should be written

Field What it says
Title and ID A short, specific name and a reference number to track it by
Severity Critical, high, medium, low or informational, usually with a CVSS score and vector
Affected assets URLs, endpoints, hosts or components, with versions
Description What the weakness is and why it exists
Evidence Requests, responses, screenshots and steps to reproduce
Impact What an attacker could do with it, in this system
Remediation A specific fix, not just "validate input"
References CVE, CWE or OWASP identifiers where they apply

Penetration test report template

Use this outline in your statement of work, so the report you receive has everything you need:

  1. Cover page: client, tester, dates, version and confidentiality marking
  2. Executive summary: overall rating, top three risks, recommendation
  3. Scope: in-scope targets with versions, out-of-scope items, test accounts, source IP addresses
  4. Methodology: method followed, manual and automated share, AI tools used and how their findings were verified
  5. Findings summary table: ID, title, severity, status
  6. Detailed findings: one entry per finding with the fields above
  7. Attack narratives
  8. Remediation plan by priority
  9. Retest results with dates
  10. Appendix A: tools and raw output
  11. Appendix B: software components in scope, from the release's SBOM

Appendix B is the one most reports lack. Without it, nobody can tell later whether a finding in a library still applies to the release you ship today.

How to read penetration test results

  • Severity is the tester's view. Re-rate each finding for your context: a high finding behind a VPN may matter less than a medium one on the login page.
  • Look at the chains. Three low findings that chain into account takeover are one critical problem.
  • Separate known from new. Outdated libraries with public CVEs are things your scanners should have caught before the test. Logic and access control flaws are what the pentest is for.
  • Check the scope. A clean report on a narrow scope says nothing about what was left out.

AI pentest reports

AI pentesting tools produce reports in minutes, and good ones attach the request that proved each finding. Hold them to the same standard: every finding needs reproducible evidence, and the report must say what was in scope and what the tool could not test. Have a person confirm critical findings before they reach a customer or an auditor.

After the report

  1. Assign every finding an owner and a deadline by severity.
  2. Fix, then retest, and add the retest results to the report.
  3. Share an attestation letter, a short summary signed by the tester, with customers who ask, and keep the full report internal.
  4. Feed recurring findings back into code review and scanning, so the next test finds new things.

Keeping the component list current

A report's list of components describes one release. CRACI does not run penetration tests, but it records every package each of your GitHub Actions builds pulls in, including the hidden dependencies that no lockfile lists, and exports the SBOM as CycloneDX or SPDX. Attach it to the report as Appendix B, and CRACI keeps checking each release's SBOM as new vulnerabilities are published, so a library finding can be traced to every release that contains it.

Penetration test reports: frequently asked questions

What should a penetration test report include?

An executive summary, the scope and method, each finding with severity, evidence, impact and a fix, attack narratives, a remediation plan and the retest results.

Who should see the pentest report?

The full report stays with the security and engineering teams, because it is a guide to attacking you. Customers usually get an attestation letter or a summary instead.

What is a pentest attestation letter?

A short letter from the testing firm stating that a test took place, its scope and dates, and the overall outcome, without the details of the findings.

How long is a typical penetration test report?

It depends on the scope and the number of findings: a small application produces a short report, a large network test a long one. Length matters less than whether each finding can be reproduced and fixed.

Can an AI pentest report be used for compliance?

Check with your auditor or customer first. Many still expect a human-led test, but AI results with reproducible evidence can support the testing between yearly tests.

Attach a component list that stays current

Record an SBOM of every release in your GitHub Actions builds, and see which releases each new vulnerability affects.

Book a demo