Choose tests based on credible failure modes
Begin with the system's threat assessment, functions and data flows. Consider code review, automated analysis, dependency checks, configuration inspection and relevant dynamic tests.
No single automated scanner establishes that an application is secure. Each technique exposes different classes of weakness, and findings need technical interpretation.
Test authentication and access controls
Verify that users and service identities can access only authorised functions and data. Include privilege escalation, session handling and unsafe defaults, particularly in externally accessible or administrative interfaces.
Authorisation problems can be missed by testing only ordinary user journeys. Construct negative test cases that attempt operations the caller should be unable to perform.
Check input handling and integrations
Test how software treats untrusted input, external services, files and messages. Review sensitive information leakage and the security assumptions made about upstream and downstream systems.
Integration tests should consider unavailable or compromised dependencies where their failure could affect covered services.
Control test environments and sensitive data
Separate test privileges from production, limit use of real sensitive information and ensure test tools do not unintentionally disrupt live systems. Preserve reproducible evidence for material defects.
Where a live assessment is necessary, agree scope, authorisation and operational safeguards before running potentially disruptive tests.
Triage and verify findings
Assess severity using exposure, exploitability, affected service and available mitigations. Assign fixes to owners and retest material issues before release where proportionate.
Track accepted exceptions separately from resolved vulnerabilities. The ability to reproduce and verify a fix is more informative than merely closing a ticket.
Use results to improve the lifecycle
Review recurrent defect patterns and release escapes, then improve design standards, developer guidance and testing coverage. Record what was tested and what remained outside scope.
Specified provider categories under Implementing Regulation (EU) 2024/2690 face detailed requirements. Others should rely on the Directive, applicable national law and risk-based controls.
Questions arising
Does NIS2 prescribe one secure development methodology?
No. The Directive establishes cybersecurity risk-management requirements, while specific methods should reflect the risks and additional applicable rules.
Should the development process include third-party components?
Yes. Dependencies and contracted development can affect system security and should be considered according to their role and risk.