CAS-005

Free CAS-005 practice questions

10 questions from our CAS-005 practice tests, each with the correct answer and a written explanation. Read them with the answers shown, or switch to quiz mode and answer first.

Back to the CAS-005 practice exams

Question 1 of 10, Governance, Risk, and Compliance

A regional hospital network is drafting its first enterprise risk register. The board states that it will accept residual risk on any single system only if the expected annual loss stays below $50,000, while the overall organization is willing to pursue new telehealth services even though they raise exposure. Which pair of terms correctly describes these two statements?

  1. AThe $50,000 per-system limit is risk tolerance, and the willingness to pursue telehealth despite added exposure is risk appetite Correct answer
  2. BThe $50,000 per-system limit is risk appetite, and the willingness to pursue telehealth despite added exposure is risk tolerance
  3. CThe $50,000 per-system limit is the risk threshold for transference, and the telehealth decision is a formal risk exception
  4. DThe $50,000 per-system limit is the residual risk, and the willingness to pursue telehealth despite added exposure is inherent risk

Answer and explanation: Risk appetite is the broad, strategic amount of risk an organization is willing to pursue to meet objectives, such as expanding into telehealth, while risk tolerance is the measurable, acceptable variation around that appetite, such as a per-system loss ceiling. Swapping the terms reverses their meaning. Residual and inherent risk describe risk levels before and after controls, not policy statements. A transference threshold and a risk exception are specific treatment decisions, not the board's general stance.

Question 2 of 10, Governance, Risk, and Compliance

A multinational retailer stores loyalty-program data for EU residents in a US-based cloud region. Legal counsel asks the security architect which mechanism would MOST directly provide a lawful basis for the transfer while keeping the current region. Which should the architect recommend?

  1. AObtain a SOC 2 Type II report from the cloud provider, which serves as an adequacy decision for transfers under GDPR
  2. BUse the EU-U.S. Data Privacy Framework if both parties are certified, or Standard Contractual Clauses with a transfer impact assessment Correct answer
  3. CEncrypt the loyalty data with customer-managed keys held in the US region, which takes the stored data outside the scope of GDPR entirely
  4. DPseudonymize customer names with tokens before storage in the US region, which classifies the data as anonymous under GDPR

Answer and explanation: International transfers under GDPR require a transfer mechanism such as an adequacy decision (the EU-U.S. Data Privacy Framework for certified US organizations) or Standard Contractual Clauses backed by a transfer impact assessment. Encryption is a safeguard but does not remove personal data from GDPR scope. Pseudonymized data is still personal data under GDPR, unlike truly anonymized data. A SOC 2 report is an assurance report and has no legal standing as an adequacy decision.

Question 3 of 10, Security Architecture

A manufacturer's zero trust project defines a policy decision point (PDP) and policy enforcement points (PEPs). An engineer proposes letting each PEP cache an allow decision for eight hours to reduce latency. Which concern should the architect raise FIRST?

  1. ACached decisions prevent the PEP from logging individual access requests, which breaks the audit trail required for SOC 2 Type II
  2. BCaching at the PEP violates the separation of the control plane and data plane, so the PDP would have to move into the data plane
  3. CLong-lived cached decisions undermine continuous authorization, so posture or risk changes go unenforced until the cache expires Correct answer
  4. DAn eight-hour cache would force every PEP to hold the policy engine's signing keys, which expands the attack surface of the trust algorithm

Answer and explanation: Zero trust relies on continuous, context-based evaluation; an eight-hour cached allow means a compromised device or revoked user keeps access until expiry, which is the primary concern. Caching does not move the PDP into the data plane; it is a performance and freshness trade-off. PEPs do not need the policy engine's signing keys to cache a decision. PEPs can still log every request while honoring a cached decision, so the audit trail is not inherently broken.

Question 4 of 10, Security Architecture

A company is replacing its hub-and-spoke MPLS network. Branch users mostly access SaaS applications, and the security team wants consistent web filtering, CASB, and ZTNA policy enforcement no matter where users connect from, without backhauling traffic to headquarters. Which architecture BEST meets these requirements?

  1. AA SASE service that combines SD-WAN with cloud-delivered SWG, CASB, ZTNA, and firewall-as-a-service Correct answer
  2. BA pair of high-availability next-generation firewalls at headquarters with full-tunnel always-on VPN for all users
  3. CA cloud-hosted reverse proxy in front of internal applications with SAML single sign-on for every web application
  4. DAn SD-WAN overlay at each branch with local internet breakout and a stateful firewall in every branch router

Answer and explanation: SASE converges networking (SD-WAN) and cloud-delivered security services such as SWG, CASB, ZTNA, and FWaaS, applying the same policy wherever users connect and avoiding backhaul. SD-WAN with local breakout alone lacks consistent CASB and ZTNA enforcement. Centralized firewalls with full-tunnel VPN reintroduce the backhaul the company wants to eliminate. A reverse proxy with SSO protects internal web apps but does not filter SaaS and general web traffic.

Question 5 of 10, Security Architecture

A company is moving a legacy three-tier application into containers orchestrated by Kubernetes. The security team wants to ensure that compromise of the web pods cannot be used to reach the database pods directly. Which control BEST achieves this?

  1. AKubernetes network policies that allow database ingress only from the application-tier pods' labels Correct answer
  2. BA single namespace for all tiers with role-based access control restricting who can run kubectl exec
  3. CPod security admission set to the restricted profile for the web pods to block privileged containers
  4. DA service mesh with mutual TLS enabled but with the default allow-all authorization policy left in place

Answer and explanation: Network policies microsegment pod traffic, so only labeled application pods can reach the database. RBAC controls API access by administrators, not pod-to-pod traffic. The restricted pod security profile hardens containers but does not block network paths. Mutual TLS without authorization policies authenticates traffic but still allows any meshed workload to connect.

Question 6 of 10, Security Engineering

An OAuth 2.0 public client, a single-page application, currently uses the implicit grant. A security review recommends a change. Which flow is BEST practice for this client today?

  1. AClient credentials flow with a secret
  2. BResource owner password credentials
  3. CAuthorization code flow with PKCE Correct answer
  4. DImplicit flow with shorter token life

Answer and explanation: Current OAuth best practice requires the authorization code flow with PKCE for public clients such as SPAs and mobile apps, because tokens are not exposed in the URL fragment and code interception is prevented. The password grant exposes user credentials to the client and is deprecated. The client credentials flow is for machine-to-machine with a confidential secret, which an SPA cannot protect. Shortening implicit token lifetimes does not fix token exposure.

Question 7 of 10, Security Engineering

An organization's DMARC aggregate reports show that a marketing vendor's legitimate emails sent on the company's behalf fail DMARC even though the vendor's IP is listed in the company's SPF record. The vendor signs mail with its own domain's DKIM key. What change will make these messages pass DMARC?

  1. APublish a second SPF TXT record for the company's domain that lists only the vendor's sending IP addresses
  2. BChange the DMARC policy from p=reject to p=none so receiving servers stop evaluating SPF and DKIM results for the domain's mail
  3. CIncrease the SPF record's DNS lookup limit beyond ten so the vendor's include mechanism is fully evaluated
  4. DHave the vendor sign with a DKIM key under the company's domain or use a company-domain return-path so an identifier aligns Correct answer

Answer and explanation: DMARC passes only when SPF or DKIM passes with an identifier aligned to the From domain; a vendor DKIM signature in the vendor's domain and a vendor-domain return-path both fail alignment. Using a DKIM selector under the company's domain or a company-domain envelope sender fixes alignment. The ten-lookup SPF limit cannot be raised. Changing to p=none only stops enforcement; messages still fail. Two SPF records for one domain cause a permerror.

Question 8 of 10, Security Engineering

A company stores database encryption keys in an on-premises HSM. To protect a sensitive analytics dataset in the cloud, the architect wants each object encrypted with a unique data key while the master key never leaves the HSM. Which technique is being described?

  1. AKey splitting, where the master key is divided among several custodians and recombined to encrypt each object
  2. BKey stretching, where a password is processed through many hash iterations to derive the object encryption key
  3. CEnvelope encryption, where data keys encrypt the objects and are themselves wrapped by a key-encryption key held in the HSM Correct answer
  4. DHomomorphic encryption, where the analytics service computes on ciphertext so the objects are never decrypted

Answer and explanation: Envelope encryption uses per-object data encryption keys that are wrapped by a key-encryption key that stays in the HSM, limiting master-key exposure and enabling efficient bulk encryption. Key stretching strengthens weak secrets such as passwords. Key splitting divides a key among custodians to enforce dual control, not per-object keys. Homomorphic encryption allows computation on ciphertext, which is a different objective.

Question 9 of 10, Security Operations. Choose 2.

A penetration tester finds that an internal API fetches a URL supplied in a JSON field and returns the response. The tester retrieves cloud instance metadata through it. Which mitigations BEST address this? (Choose two.)

  1. ARequire session tokens for instance metadata, such as IMDSv2, and limit the metadata hop count Correct answer
  2. BAdd a CAPTCHA to the client application before it calls the URL-fetching endpoint
  3. CIncrease the API's TLS version to 1.3 so that the fetched response cannot be intercepted
  4. DEncode the returned content with HTML entities before the API sends its response to the client
  5. EValidate destinations against an allow list of approved hosts and block link-local and private ranges Correct answer

Answer and explanation: This is server-side request forgery; allow-listing destinations and blocking internal ranges prevents the server from reaching metadata endpoints, and token-based metadata services with a hop limit block simple SSRF retrieval. Output encoding protects browsers from script injection, not server-side fetches. TLS 1.3 protects the transport but not what the server requests. CAPTCHAs throttle automation but do not stop a single malicious request.

Question 10 of 10, Security Operations

An analyst receives a suspicious executable and wants to observe its network connections, file changes, and registry modifications without risking production systems. Which approach is BEST?

  1. ARun the sample on the analyst's workstation with EDR in audit mode
  2. BUpload the sample to a public file-sharing site for team review
  3. CDetonate the sample in an isolated sandbox that records its behavior Correct answer
  4. DDisassemble the sample in a debugger and review its code paths without any execution

Answer and explanation: Detonation in an isolated sandbox safely captures runtime behavior such as network, file, and registry activity. Uploading to a public site may alert the adversary and leak data. Running malware on a corporate workstation risks infection even with EDR. Static disassembly is valuable but does not directly observe runtime behavior.