CRA cybersecurity risk assessment: what Article 13 and Annex VII require

The Cyber Resilience Act requires manufacturers to assess the cybersecurity risks of each product, document the assessment, use it throughout design, production and maintenance, and state in it how every Annex I requirement applies. It belongs in the technical documentation and is updated during the support period.

By the Vellaci teamPublished Updated 7 min readOperational guidance, not legal advice

Share:LinkedInXE-mail

Key facts

Legal basis
Art. 13(2)–(4); Annex VII point 3
Minimum content
Intended purpose, reasonably foreseeable use, conditions of use such as the operational environment and assets, and the expected time in use
What it decides
Whether and how each Annex I Part I point (2)(a)–(m) applies, and how Part I(1) and Part II are applied
Not applicable?
A clear justification in the technical documentation (Art. 13(4))
Kept current
Documented and updated as appropriate during the support period (Art. 13(3))
Other EU law
May form part of a risk assessment required by other Union legislation that also covers the product (Art. 13(4))

What the regulation requires

Article 13(2) requires manufacturers to undertake an assessment of the cybersecurity risks associated with a product with digital elements and to take its outcome into account during planning, design, development, production, delivery and maintenance, with a view to minimising cybersecurity risks, preventing incidents and minimising their impact, including on users' health and safety.

Article 13(3) sets the floor for its content: an analysis of cybersecurity risks based on the intended purpose and reasonably foreseeable use, and on the conditions of use — such as the operational environment or the assets to be protected — taking into account how long the product is expected to be in use. The assessment must indicate whether, and if so how, each security requirement in Annex I Part I point (2) applies and is implemented, and how the manufacturer applies Part I point (1) and the vulnerability-handling requirements of Part II. Article 13(4) puts it in the technical documentation and requires a clear justification for every essential requirement that does not apply.

A method that produces the evidence

The regulation prescribes content, not a method. Any structured method works if its output answers the Article 13(3) questions for a named product and version. A sequence that does:

  1. Define the product boundary: hardware, firmware, applications, the remote data processing the product depends on, and the third-party components from the SBOM.
  2. State the intended purpose and the reasonably foreseeable use, including misuse a user could plausibly attempt.
  3. Describe the conditions of use: operational environment, network exposure, users and administrators, and the assets to protect — data, functions, connected systems, safety.
  4. Record the expected time in use; it drives the support period and how long controls must remain effective.
  5. Identify threat scenarios per asset and interface, for example with STRIDE or attack trees, and rate likelihood and impact on a declared scale.
  6. Map each scenario to Annex I Part I points and to the controls that address it; mark points that do not apply and write why.
  7. Record residual risk, the decision to accept or reduce it, the approver and the date.
  8. Link the evidence: design documents, test reports, SBOM, configuration defaults.
  9. Define review triggers so the assessment changes when the product or the threat picture does.

Mapping Annex I to the assessment

Annex I Part I pointQuestion the assessment answersExample outcome
(2)(a) no known exploitable vulnerabilitiesHow do we know a release has none?Release gate on SBOM matching and triage; exceptions documented
(2)(b) secure by defaultWhich defaults could expose users?No default credentials; remote management off by default
(2)(c) security updatesHow do fixes reach every device, and how fast?Signed automatic updates, enabled by default, with postponement
(2)(d) access controlWho can do what, and how is abuse detected?Per-device credentials, role separation, failed-login reporting
(2)(e)–(f) confidentiality and integrityWhich data and commands need protection, at rest and in transit?TLS for all traffic; signed configuration; encrypted secrets
(2)(g) data minimisationWhich data does each function really need?Telemetry limited to fields required by the stated purpose
(2)(h)–(i) availabilityWhat keeps working under attack, and what harms other networks?Local fallback mode; outbound rate limits
(2)(m) data removalCan users wipe the device securely?Factory reset erases keys and user data; tested per release

Worked example: a connected thermostat

An illustrative assessment for a Wi-Fi thermostat with a mobile app and a cloud service it depends on for remote control. The cloud service is remote data processing and is inside the assessment boundary.

ScenarioAnnex I pointsControlResidual risk
Account takeover through reused passwords lets an attacker change heating schedules(2)(d)Multi-factor authentication option, login rate limiting, new-device alertsLow — accepted, reviewed yearly
Tampered firmware pushed through a compromised update channel(2)(c), (2)(f)Signed firmware verified on device; signing keys in a hardware security moduleLow
Local network attacker reads Wi-Fi credentials from the device(2)(e)Credentials encrypted in a secure storage areaMedium — hardware revision planned
Cloud outage leaves homes without heating control(2)(h)Local schedule continues offline; physical buttons always workLow
Telemetry reveals when a home is empty(2)(e), (2)(g)Occupancy data aggregated; not retained beyond 30 daysLow
Logging of user activity(2)(l)Security events logged with user opt-out as requiredLow

Standards and methods you can reuse

ISO/IEC 27005 gives a general information-security risk process; IEC 62443-3-2 covers security risk assessment for industrial system design; threat-modelling techniques such as STRIDE and attack trees produce the scenario list. For the CRA specifically, CEN and CENELEC are finalising EN 40000-1-2, 'Principles, product risk management, and lifecycle activities'; its CEN record showed ratification scheduled for 2 November 2026 and no Official Journal citation expected, so it will be a strong reference for method but not by itself a presumption of conformity.

Products also covered by other EU legislation — machinery, radio equipment, medical-adjacent devices — may fold the cybersecurity assessment into the risk assessment that legislation requires (Article 13(4)), provided the CRA content is visible in it.

Keeping it current

Article 13(3) requires the assessment to be updated as appropriate during the support period, and Article 13(7) links documented vulnerabilities to updates of it. Practical triggers:

  • A new release, and in particular any change that could be a substantial modification.
  • An actively exploited vulnerability or a severe incident affecting the product or a similar one.
  • A new third-party component, or an existing one reaching end of life.
  • A change in intended purpose, market, user group or operational environment.
  • A scheduled review — at least yearly for products with long support periods.

Common mistakes

  • A generic company-level assessment with no product, version or boundary.
  • Leaving out the cloud back-end the product needs to work.
  • Marking Annex I points 'not applicable' without a reason.
  • No owner, approver or date — and no update after an incident.
  • Risk ratings with no link to the controls and test evidence that justify them.

Frequently asked questions

Does the CRA prescribe a risk assessment method?
No. Article 13(3) prescribes what the assessment must cover and conclude, not how to perform it. Use a structured method you can repeat — ISO/IEC 27005, IEC 62443-3-2 or threat modelling — and make sure the output addresses every Annex I point for the specific product.
Is the CRA risk assessment the same as a DPIA or an ISO 27001 risk assessment?
No. A DPIA assesses risks to individuals from processing personal data, and an ISO 27001 assessment covers an organisation's information security. The CRA assessment is about a product's cybersecurity across its lifecycle. The inputs overlap, but the CRA one must be product-specific and map to Annex I.
How often must the CRA risk assessment be updated?
As appropriate during the support period. In practice: with every release that changes the product's risk, after an exploited vulnerability or severe incident, when components or the use context change, and at a fixed review interval.
Can one risk assessment cover a whole product family?
A shared assessment can cover common platform elements, but each product placed on the market needs its own conclusions for its intended purpose, environment and Annex I applicability. Keep the shared part referenced and the product-specific part explicit.
Who should approve the risk assessment?
The regulation does not name a role. Choose someone accountable for the product's security — typically the product security owner — record the approval and date, and involve engineering and compliance in the review.

From requirement to evidence

What the obligations in this guide mean for an engineering, security or compliance team, the Vellaci workflow that supports each one, and the records it leaves. The full map for every CRA obligation is on the CRA compliance software page.

How to read this. Official requirement summarises the text of Regulation (EU) 2024/2847 with its reference. What your team does is Vellaci’s operational interpretation, not legal advice. Vellaci workflow and Evidence describe what the product does today. Methodology · reviewed 2026-10-05.

  1. Cybersecurity risk assessment

    Article 13(2)–(3)Applies from 11 Dec 2027
    Official requirement
    Undertake an assessment of the cybersecurity risks of the product, take its outcome into account during planning, design, development, production, delivery and maintenance, document it and update it as appropriate during the support period.
    What your team does (interpretation)
    Write the risk assessment per product, link it to the Annex I controls it justifies, and revisit it when vulnerabilities or incidents change the picture.
    Vellaci workflow
    Risk assessment section in the per-product technical documentation workspace, with version history; per-product obligations view. How it works
    Evidence produced
    • Versioned risk assessment section with author and status
    • Link between findings and the assessment they updated
  2. Technical documentation

    Article 13(12)–(13); Article 31; Annex VIIApplies from 11 Dec 2027
    Official requirement
    Draw up the technical documentation before placing the product on the market, containing at least the elements in Annex VII, and keep it, with the EU declaration of conformity, at the disposal of market surveillance authorities for at least ten years after placing on the market or for the support period, whichever is longer.
    What your team does (interpretation)
    Assemble the Annex VII file progressively from records you already keep — SBOMs, vulnerability handling, risk assessment, support period, test reports — instead of writing it once before an audit.
    Vellaci workflow
    Fourteen Annex VII-oriented sections per product with status, version history and evidence links, exported as a ZIP technical package. How it works
    Evidence produced
    • Section versions with author and approval status
    • ZIP technical package (product record, SBOMs, sections, evidence index)

Draft the risk assessment worksheet now.

Document use, threat scenarios, controls, residual risk and evidence for one product in the free worksheet, then keep it with the technical documentation.