Skip to content
Tutorials Center
My learning

Topics

Lesson 44 of 53 ยท Reporting and Careers

Writing a Penetration Test Report

Learn how to write a penetration test report: executive summary, findings structure, severity ratings, reproduction steps, remediation advice and the debrief.

A penetration test is only as valuable as its report. The client does not pay for the clever thing you did on Tuesday. They pay for a document that tells leaders what the risk is, tells engineers exactly what to fix, and can be filed as proof that a test took place. Technically strong testers who write vague reports leave their clients with nothing to act on.

This lesson covers the structure of a good report, how to describe a finding, how to rate severity honestly, how to handle evidence safely, and how to run the debrief. Real testing tools cannot run here, so report samples are read-only text. The runnable part is a JavaScript stand-in: a small program that turns structured findings data into a report. It illustrates why structured notes make reporting faster, and it is not a commercial reporting tool. For where reporting fits in the work, see The Penetration Testing Process.

Two readers, one report

Leaders Want risk, impact and priorities.
Engineers Want steps to reproduce and fix.

Write the summary for the first reader and the findings for the second.

1 / 2

The shape of a report

Most reports follow the same outline, so readers know where to look:

  1. Title page and document control. Client, test dates, version, authors, confidentiality marking, and who may read it.
  2. Executive summary. One to two pages, no jargon. State what was tested and why, the overall risk in plain words, the top three to five issues in business terms, and the most important next actions.
  3. Scope and method. What was in scope (systems, addresses, time window), what was excluded, the approach (black, grey or white box), the tools in general terms, and any limits such as "no production data was modified" or "social engineering was out of scope". This ties back to the rules of engagement.
  4. Findings summary. A table of every finding with ID, title, severity and affected asset, sorted by severity.
  5. Detailed findings. One section per finding (below).
  6. Positive observations. What was done well. It is accurate, it motivates the team, and it tells them what not to break.
  7. Appendices. Raw scan output, tool versions, a glossary, and the severity scale.

Writing the executive summary

Write it last, but put it first. A weak summary lists techniques ("we found three XSS issues"). A strong one states the consequence ("an attacker with a free account could read other customers' invoices"). Use plain sentences and one honest overall judgment, such as "The external perimeter is well protected, but the customer portal has weaknesses that allow unauthorized access to customer data." Avoid fear and avoid sugar-coating.

Anatomy of a finding

Every finding should answer six questions, in a consistent template:

  • Title. Short and specific: "Customer invoices are accessible without ownership checks", not "IDOR".
  • Severity. Critical, high, medium, low or informational, with the reasoning.
  • Affected assets. Exact hosts, URLs, parameters, versions.
  • Description. What is wrong and why, in a few paragraphs.
  • Impact. What an attacker could achieve in this environment, for example "read any customer's invoice by changing the number in the URL".
  • Steps to reproduce. Numbered, exact and minimal, so a developer who was not there can see it for themselves.
  • Evidence. Screenshots, request and response excerpts, or log lines, with sensitive data redacted.
  • Remediation. What to change, in order of preference, and how to verify the fix.
  • References. Relevant standards or advisories (a CWE number, an OWASP category).

Here is a sample (made-up) detailed finding in plain text:

TEXT
F-02  Customer invoices readable without ownership check        HIGH

Affected:  https://portal.example.test/invoices/<id>
Description:
  The invoice endpoint requires a login but does not check that the
  invoice belongs to the logged-in customer.
Impact:
  Any registered customer can read every other customer's invoices,
  including names, addresses and amounts. IDs are sequential, so the
  full set can be enumerated.
Steps to reproduce:
  1. Log in as test user A and open /invoices/1041 (your own invoice).
  2. Change the number to 1042 (belongs to test user B).
  3. Observe that user B's invoice is returned with status 200.
Remediation:
  Authorize each request against the invoice owner on the server.
  Return 404 for records the caller does not own. Add an automated test.
References: CWE-639, OWASP API1 Broken Object Level Authorization

Notice the reproduction uses test accounts created for the engagement. Never prove impact by reading real customers' data beyond the minimum needed to show the flaw, and stop there.

Rating severity honestly

Severity combines how easy a flaw is to exploit with how bad the result is, in this client's context. Many teams start from the CVSS base score (Vulnerability Scanning and Triage) and adjust for environment. A simple scale you can explain in an appendix:

Severity Typical meaning
Critical Easy to exploit, remote and leads to full compromise or mass data exposure
High Serious impact, or moderate effort to reach it
Medium Needs special conditions or limited impact
Low Small impact, hard to exploit, or defense-in-depth gap
Informational No direct risk, but worth knowing

Three habits keep ratings credible. Explain the rating in one sentence. Do not inflate to look impressive; clients stop believing a report where everything is critical. Consider chains: two mediums that combine into a path to administrator may deserve a high rating on their own, and you can add a short "attack narrative" showing the path.

Handling evidence and the report itself

A report describes how to get into the client's systems, so it is highly sensitive. Mark it confidential, deliver it through an encrypted channel or portal rather than ordinary email, record who received it, and agree on how long you keep test data (and delete it on schedule). Redact passwords, tokens and personal data from screenshots and logs. Keep working notes organized while you test, with timestamps, because reproducing a finding days later from memory is slow and error-prone. Evidence habits such as hashing files are covered in Incident Response and Digital Forensics Basics.

The debrief and retest

Walk the client through the results live, with both leaders and engineers present when you can. Explain the top findings, answer questions, and agree on a remediation order: fix cheap, high-impact items first. Offer a retest after the client fixes the issues, and record the outcome per finding as fixed, partly fixed or still open. A short retest letter is often what auditors want to see.

Stand-in: generate a report from structured data

Typing a report by hand is slow and inconsistent. Many teams keep findings as structured data and generate the document from templates. This stand-in does that for a small made-up engagement: it sorts findings by severity, builds the summary table, counts by severity and prints each detailed finding.

JS Findings data to report

Press Run (or Ctrl + Enter) to see the console output here.

No output.

What is real? Keeping findings as data and generating tables, counts and detail sections is how real reporting tools and templates work. The three findings and the plain-text output are the stand-in. Real reports add branding, evidence images and approval workflow.

Stand-in: a finding quality checker

A quick review catches vague findings before the client sees them. This checker flags findings with missing parts.

JS Lint your findings

Press Run (or Ctrl + Enter) to see the console output here.

No output.

Risk assessment, remediation planning and report types

A finding list becomes useful when it is turned into decisions: what to fix first, who fixes it and by when. That is risk assessment followed by remediation planning.

Rating risk

Risk is usually described as likelihood ร— impact. Likelihood depends on how easy the flaw is to exploit (is it reachable from the internet? does it need credentials? is there a public exploit?) and how motivated an attacker might be. Impact depends on what the system holds and does: personal data, money, safety, reputation, regulatory exposure. A moderate technical flaw on a system that holds customer payment data can outrank a severe flaw on an isolated test server. Scoring systems such as CVSS give a technical base score, but the business context adjusts it; the severity guidance earlier in this lesson explains how to stay honest about it.

A simple rating matrix helps keep ratings consistent across testers:

Likelihood \ Impact Low Medium High
High Medium High Critical
Medium Low Medium High
Low Low Low Medium

Remediation plans

For each finding the plan answers five questions: what is the fix (specific and testable), who is the owner, what is the priority and due date (tied to the rating: for example critical in days, high in weeks), what interim mitigation exists if the fix takes time (a firewall rule, a feature flag, extra monitoring), and how will it be verified (retest by the testers or an automated check). Group related findings: one outdated library may explain ten scanner results. Track the plan in a ticket system rather than only in a PDF, because the report is a snapshot and the work happens afterwards (Vulnerability Tracker project).

Vulnerability assessment report versus penetration test report

The two documents answer different questions:

  • A vulnerability assessment report is broad: it lists as many weaknesses as were found across many systems, usually from automated scans plus review, ranked by severity, with little or no exploitation. It answers "what is wrong across the estate?"
  • A penetration test report is deep: it shows which weaknesses could be chained together by an attacker, with proof of exploitation, the impact reached and the attack path. It answers "what could a determined attacker actually do?"

Both need an executive summary, scope, method, findings with evidence and remediation advice. The stand-in below computes priority and due dates from likelihood and impact, producing a remediation plan in order. The data is invented.

Node.js Turn findings into a prioritized plan

Press Run (or Ctrl + Enter) to see the result here.

Program input ยท what you'd type while the program runs

Type the answers before you press Run, one per line. Each time the program reads a line (with node:readline or process.stdin), it gets the next line from this box.

The sorted output is exactly what a project manager needs, and the counts feed the executive summary. Always arrange a retest at the end: a finding is closed only when someone has checked that the fix works, which is what turns a report into measurable improvement.

Check your understanding

Quick check ยท 4 questions

of 4 answered You scored / 4 ยท perfect!

  1. 1. Who is the executive summary mainly written for?

    Leaders need the overall risk, the consequences and the next actions. Technical detail belongs in the findings section.
  2. 2. Which finding title is best?

    A good title names the flaw and where it is, so readers understand it before opening the details.
  3. 3. Why should reproduction steps be numbered and exact?

    Reproducible steps let developers verify the problem and testers verify the fix during a retest.
  4. 4. How should the finished report be treated?

    A report is a map of how to attack the client. Protect it and control distribution.

Summary

  • A report has two audiences: leaders (executive summary) and engineers (detailed findings).
  • Each finding needs a specific title, severity with reasoning, affected assets, impact, exact reproduction steps, evidence, remediation and references.
  • Rate severity honestly, in the client's context, and watch for chains of lower issues.
  • Protect the report and the evidence, redact sensitive data, and use test accounts.
  • Debrief live, agree a fix order, and retest.

Next: Certifications, Practice Platforms and Careers.