Identify security responsibilities at initiation
Before development starts, determine who owns the system, which covered services it supports and who can accept residual risks. Define expectations for externally supplied components and sensitive data.
Security responsibilities should follow the project across teams rather than disappearing when requirements pass from procurement to engineering.
Build requirements into design and architecture
Set criteria for authentication, authorisation, protection of information, monitoring, recoverability and secure interfaces. Assess threats arising from privileged functions, trust relationships and external integrations.
Record significant tradeoffs where the chosen architecture cannot meet a preferred control. A risk owner should understand the consequences before accepting an exception.
Apply controls in implementation
Use appropriate peer review, access protection, secrets management and dependency oversight. Protect development environments and pipelines from changes that could alter production releases without authorisation.
Avoid treating a source-code scan as a complete representation of software security. Configuration, infrastructure and release processes affect the deployed system.
Verify and approve releases
Plan proportionate security tests for critical functions, review findings and confirm that material defects are addressed or appropriately accepted. Establish a traceable link between approved code, built artefacts and released versions.
Testing should reflect the threat and service impact. A low-risk content update does not necessarily need the same scrutiny as a change to payment or identity logic.
Maintain and retire securely
After deployment, monitor vulnerabilities, support levels, configuration changes and new dependencies. Establish processes for emergency fixes and safe rollback, including service continuity constraints.
At retirement, remove unnecessary identities, integrations and data copies while preserving records needed under applicable law and contractual duties.
Evaluate lifecycle effectiveness
Track recurring security failures, overdue remediation, unsupported components and exceptions. Retain review and test evidence that shows what was checked and what remains unresolved.
Check Member State law and any provider-specific provisions under Implementing Regulation 2024/2690 before stating a particular lifecycle control is legally mandatory.
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.