Methodology

Authorisation first. Evidence always.

A test is only useful if its results can be trusted, and only legal if it was agreed in writing beforehand. This page describes exactly how an engagement runs — including what I refuse to do.

The engagement, step by step

  1. Scoping

    A short call plus a written scope list: applications, API endpoints, mobile builds, admin areas, environments, test accounts, deadlines and anything you explicitly want untouched. I also ask what the system does with money and personal data, because that drives severity.

  2. Written authorisation

    The document you sign names: the assets in scope, the test window and testing hours, my source IP, the test accounts, the contact person who can stop the test, and the data-handling terms. No signature, no packets.

  3. Reconnaissance

    Subdomains, DNS and certificate records, HTTP headers, publicly served files, client-side bundles, mobile app packages and app-store metadata. Anything publicly readable is fair game and needs no guesswork — this is also where exposed keys and secrets usually appear.

  4. Active testing

    Rate-limited scanning and manual testing on the in-scope assets only: authentication and session handling, OTP flows, authorisation between accounts (IDOR / BOLA), injection classes, file upload and handling, business-logic abuse, TLS and header hardening, cloud and storage rules.

  5. Impact demonstration

    Where a finding is exploitable, I take it just far enough to prove impact — the smallest possible proof of concept, on your own test data. No persistence, no backdoors, no exfiltration, no changes to production data.

  6. Reporting

    Critical findings are reported within 24 hours of confirmation, with a workaround. The full report follows within 10 working days: owner-facing summary plus a developer-facing task list, with evidence and reproduction steps per finding.

  7. Retest and closure

    Once fixes are deployed I retest them and deliver a closing statement: closed, partially closed, still open, plus anything the fix introduced. The engagement ends with a written record of the state of the system on that date.

Rules of engagement

The same rules apply to every engagement. Deviations are possible, but only in writing, signed by both sides before the work starts.

AreaAlwaysNever without a separate written addendum
Scope Only assets listed and signed for; one environment at a time Other domains, subdomains, customers' systems, anything hosted by a third party
Timing Agreed window and working hours; notice before any unusual burst of traffic Testing during your blackout periods, launches or peak trading hours
Load Rate limits and throttling; a halt on request, immediately, verbally or in writing Denial-of-service, stress or load testing, brute force against production accounts
Data Your test accounts; personal data masked in reports; evidence encrypted at rest Reading, exporting, modifying or deleting real customer data
Credentials Only accounts and keys you provide for the test Using keys found in public code — those are reported as findings, never exercised
Access Read, probe, demonstrate Persistence, backdoors, webshells, test accounts or files left behind
Third parties List every third-party service in the report so you can notify them Testing payment providers, cloud providers, chat widgets, hosting or app stores
Disclosure Confidential handling; findings shared only with the people you name Publishing anything before the agreed disclosure date, or without written consent

Legal basis and data handling

Why the signature matters

Under Dutch law, accessing a computer system without the owner's permission is a criminal offence (article 138ab of the Dutch Criminal Code, and related provisions on data interception and damage). A security review does not create that permission by itself.

The signed authorisation is therefore not paperwork — it is the legal basis for the work. It records who may test, what may be tested, when, from where, and who can stop it. It also states that the client will not report activities that stay inside the agreed scope.

If a test uncovers something outside the agreed scope — a third-party service, another tenant, or personal data that is not needed to prove a finding — I stop, record what I saw, and tell you. I do not continue on my own initiative.

Personal data (GDPR / AVG)

  • You remain the data controller; data is processed only to deliver the test and the report.
  • Data minimisation: I collect only what is needed to prove a finding, and take the smallest possible proof of concept.
  • Masking: names, e-mail addresses, phone numbers and payment identifiers are truncated or hashed in every report and screenshot that leaves my system.
  • Retention: evidence is stored encrypted and deleted 90 days after delivery, unless you ask for it to be kept longer in writing.
  • Breach: if data is exposed by accident, you are notified within 24 hours, with the scope and the mitigation.

Severity scale and reporting timeline

How severity is decided

RatingWhat it normally means
CriticalPayment, identity or customer data reachable by an unauthenticated attacker; account takeover at scale
HighPrivilege escalation, access to another customer's data, bypass of a core control
MediumRequires a precondition (a user action, a specific role) but leads to real impact
LowHardening, information disclosure, defence in depth

Every rating carries a one-line justification and a confidence tag (confirmed / likely / possible), so a low-confidence finding is never presented as a certainty.

Timeline

  • Immediately: stop and call you if a test risks availability or exposes data in real time.
  • Within 24 hours: critical findings confirmed during the test, with a temporary workaround.
  • Within 10 working days: complete report with evidence and priority-ordered fixes.
  • Within 30 days of your fix: retest and closing statement.
  • 90 days after the first report: earliest date for any responsible-disclosure discussion, and only with your consent.

What I need from you before starting

Checklist

  • A decision maker who can sign the authorisation
  • The asset list: domains, subdomains, API base URLs, mobile builds, admin areas
  • Two test accounts (a normal user and an admin) — for authorisation testing
  • A staging copy where available; if not, agreed testing hours on production
  • A phone number for emergencies and a maintenance/blackout calendar
  • Access to source code and configuration if you also want a code review

Then, in return

  • A fixed price and a delivery date, in writing, before any work starts
  • Findings reported as they are confirmed, not held until the end
  • Plain-language summary for the business owner, task list for the developer
  • Retest of the fixes within 30 days, included in the engagement
  • Evidence you can archive — and hand to an auditor, an insurer or your client