Anthropic CCAR-P (Claude Certified Architect - Professional) Practice Test 18

Question 170

An employee asks for a specific error code present verbatim in one troubleshooting page; vector search misses it. Which next step is most defensible?

  1. Use hybrid lexical-semantic retrieval with reranking tested on both query families.
  2. Use semantic vector retrieval with an evaluated embedding model.
  3. Filter on effective jurisdiction and date before semantic ranking.
  4. Add lexical or hybrid search for exact identifiers alongside semantic retrieval.

Explanation

Correct Answer

D:

Sparse keyword matching handles rare codes that dense similarity can underweight. Build query cases that verify exact-token retrieval for identifiers and check whether the chosen method returns complete, applicable evidence rather than merely similar text.

Incorrect Answers

A:

The design concern is real but different: mixed identifier-and-semantic search. It does not impose the specific requirement for exact-token retrieval for identifiers; a suitable solution must add lexical or hybrid search for exact identifiers alongside semantic retrieval and demonstrate the result.

B:

This addresses paraphrase-aware document retrieval; however, the problem asks for exact-token retrieval for identifiers. Without the safeguard to add lexical or hybrid search for exact identifiers alongside semantic retrieval, the original failure mode is still possible.

C:

That choice would be responsive to applicability-first document filtering, but the reported defect is different. It leaves exact-token retrieval for identifiers unproved because there is no commitment to add lexical or hybrid search for exact identifiers alongside semantic retrieval.

Question 171

A user paraphrases a complicated policy in everyday language with no shared keywords. Which approach should be recommended?

  1. Use hybrid lexical-semantic retrieval with reranking tested on both query families.
  2. Use semantic vector retrieval with an evaluated embedding model.
  3. Add lexical or hybrid search for exact identifiers alongside semantic retrieval.
  4. Use hierarchical retrieval that includes referenced sections and surrounding definitions.

Explanation

Correct Answer

B:

Dense similarity can connect phrasing variants when exact-term matching has poor recall. Build query cases that verify paraphrase-aware document retrieval and check whether the chosen method returns complete, applicable evidence rather than merely similar text.

Incorrect Answers

A:

This recommendation targets mixed identifier-and-semantic search at a different point in the workflow. It leaves paraphrase-aware document retrieval outside the design contract and does not require the team to use semantic vector retrieval with an evaluated embedding model.

C:

This would make sense as an additional project activity for exact-token retrieval for identifiers. The priority required here is paraphrase-aware document retrieval, so the workflow still needs a control that will use semantic vector retrieval with an evaluated embedding model.

D:

By itself, this improves dependency-aware retrieval across clauses, but the case is not asking the architect to optimize that dimension. The requirement is paraphrase-aware document retrieval, supported by an approach to use semantic vector retrieval with an evaluated embedding model.

Question 172

A question asks for the current order status, which changes minute by minute in a relational database. What is the best production design response?

  1. Use a paginated database query with completeness checks.
  2. Use a controlled SQL or analytics query and verify the computation before natural-language explanation.
  3. Query an authorized live transactional API rather than relying on a periodically embedded snapshot.
  4. Select authoritative current versions and surface unresolved conflicts with citations.

Explanation

Correct Answer

C:

Operational state is better read from the system of record than a lagging vector index. Build query cases that verify fresh structured record lookup and check whether the chosen method returns complete, applicable evidence rather than merely similar text.

Incorrect Answers

A:

This shifts attention toward complete entity-set retrieval when the relevant decision is about fresh structured record lookup. Unless the architecture can query an authorized live transactional API rather than relying on a periodically embedded snapshot, the failure described in the stem remains unresolved.

B:

For a workflow whose primary weakness was exact numerical aggregation from structured data, that proposal could be reasonable. Here the unmet obligation is fresh structured record lookup, and that requires the system to query an authorized live transactional API rather than relying on a periodically embedded snapshot.

D:

It is aimed at a different outcome, version-aware authoritative evidence. No evidence is given that it provides fresh structured record lookup; the needed mechanism is to query an authorized live transactional API rather than relying on a periodically embedded snapshot at the appropriate boundary.

Question 173

An analyst asks how many suppliers meet several numeric thresholds across thousands of rows. Which change most directly satisfies this requirement?

  1. Use a paginated database query with completeness checks.
  2. Use semantic vector retrieval with an evaluated embedding model.
  3. Use a controlled SQL or analytics query and verify the computation before natural-language explanation.
  4. Query an authorized live transactional API rather than relying on a periodically embedded snapshot.

Explanation

Correct Answer

C:

Aggregation is deterministic structured-data work; searching text chunks cannot guarantee complete counts. Build query cases that verify exact numerical aggregation from structured data and check whether the chosen method returns complete, applicable evidence rather than merely similar text.

Incorrect Answers

A:

There is a legitimate use case for complete entity-set retrieval, but it does not supply the test or enforcement needed for exact numerical aggregation from structured data. The design needs to use a controlled SQL or analytics query and verify the computation before natural-language explanation.

B:

Consider its actual operating scope: paraphrase-aware document retrieval. The requested assurance is exact numerical aggregation from structured data, which calls for an implementation that can use a controlled SQL or analytics query and verify the computation before natural-language explanation, not merely for a related improvement.

D:

The proposed measure is aimed at fresh structured record lookup, not exact numerical aggregation from structured data. The architecture must instead use a controlled SQL or analytics query and verify the computation before natural-language explanation; a separate change is still needed for exact numerical aggregation from structured data.

Question 174

A support knowledge base combines rare part numbers with descriptive troubleshooting narratives. Which solution best resolves this operational gap?

  1. Use semantic vector retrieval with an evaluated embedding model.
  2. Use hybrid lexical-semantic retrieval with reranking tested on both query families.
  3. Add lexical or hybrid search for exact identifiers alongside semantic retrieval.
  4. Use a paginated database query with completeness checks.

Explanation

Correct Answer

B:

Complementary search signals cover identifiers and conceptual questions without sacrificing one class. Build query cases that verify mixed identifier-and-semantic search and check whether the chosen method returns complete, applicable evidence rather than merely similar text.

Incorrect Answers

A:

By itself, this improves paraphrase-aware document retrieval, but the case is not asking the architect to optimize that dimension. The requirement is mixed identifier-and-semantic search, supported by an approach to use hybrid lexical-semantic retrieval with reranking tested on both query families.

C:

The option might improve exact-token retrieval for identifiers; that remains insufficient for mixed identifier-and-semantic search. The case is resolved by mixed identifier-and-semantic search, which depends on a design that will use hybrid lexical-semantic retrieval with reranking tested on both query families.

D:

The proposed action prioritizes complete entity-set retrieval. That can coexist with the right solution, but it cannot replace mixed identifier-and-semantic search: the application must still use hybrid lexical-semantic retrieval with reranking tested on both query families.

Question 175

A contract question refers to an exception defined in another section far from the main clause. What design decision would best fit these facts?

  1. Use a paginated database query with completeness checks.
  2. Use hierarchical retrieval that includes referenced sections and surrounding definitions.
  3. Use hybrid lexical-semantic retrieval with reranking tested on both query families.
  4. Use semantic vector retrieval with an evaluated embedding model.
  5. Select authoritative current versions and surface unresolved conflicts with citations.

Explanation

Correct Answer

B:

Cross-references require related context that a single isolated chunk does not contain. Build query cases that verify dependency-aware retrieval across clauses and check whether the chosen method returns complete, applicable evidence rather than merely similar text.

Incorrect Answers

A:

It is aimed at a different outcome, complete entity-set retrieval. No evidence is given that it provides dependency-aware retrieval across clauses; the needed mechanism is to use hierarchical retrieval that includes referenced sections and surrounding definitions at the appropriate boundary.

C:

That is primarily a technique for mixed identifier-and-semantic search. It lacks the explicit measure for dependency-aware retrieval across clauses that is necessary here, namely a system that will use hierarchical retrieval that includes referenced sections and surrounding definitions.

D:

Improving paraphrase-aware document retrieval does not imply dependency-aware retrieval across clauses. The selected design must use hierarchical retrieval that includes referenced sections and surrounding definitions, whereas this choice omits that direct protection or verification step.

E:

This recommendation targets version-aware authoritative evidence at a different point in the workflow. It leaves dependency-aware retrieval across clauses outside the design contract and does not require the team to use hierarchical retrieval that includes referenced sections and surrounding definitions.

Question 176

An application receives location-specific policy requests where documents share almost identical language. What should the solution architect prioritize?

  1. Use a paginated database query with completeness checks.
  2. Use semantic vector retrieval with an evaluated embedding model.
  3. Add lexical or hybrid search for exact identifiers alongside semantic retrieval.
  4. Filter on effective jurisdiction and date before semantic ranking.

Explanation

Correct Answer

D:

Applicability metadata is a hard constraint; similarity alone may retrieve the wrong law. Build query cases that verify applicability-first document filtering and check whether the chosen method returns complete, applicable evidence rather than merely similar text.

Incorrect Answers

A:

The proposed measure is aimed at complete entity-set retrieval, not applicability-first document filtering. The architecture must instead filter on effective jurisdiction and date before semantic ranking; a separate change is still needed for applicability-first document filtering.

B:

The relevant trade-off is not primarily paraphrase-aware document retrieval. A successful fix needs applicability-first document filtering; otherwise applicability-first document filtering remains unverified even after this change.

C:

The control would be justified if exact-token retrieval for identifiers were the limiting factor. In this scenario the decisive target is applicability-first document filtering; success requires an architecture that will filter on effective jurisdiction and date before semantic ranking.

Question 177

A RAG assistant answers a query from three conflicting revisions without indicating their dates. Which design choice best matches the stated constraint?

  1. Use a paginated database query with completeness checks.
  2. Use hierarchical retrieval that includes referenced sections and surrounding definitions.
  3. Use hybrid lexical-semantic retrieval with reranking tested on both query families.
  4. Select authoritative current versions and surface unresolved conflicts with citations.

Explanation

Correct Answer

D:

Retrieval should reconcile validity periods, not blend mutually inconsistent source statements. Build query cases that verify version-aware authoritative evidence and check whether the chosen method returns complete, applicable evidence rather than merely similar text.

Incorrect Answers

A:

This addresses complete entity-set retrieval; however, the problem asks for version-aware authoritative evidence. Without the safeguard to select authoritative current versions and surface unresolved conflicts with citations, the original failure mode is still possible.

B:

The design concern is real but different: dependency-aware retrieval across clauses. It does not impose the specific requirement for version-aware authoritative evidence; a suitable solution must select authoritative current versions and surface unresolved conflicts with citations and demonstrate the result.

C:

The proposed measure is aimed at mixed identifier-and-semantic search, not version-aware authoritative evidence. The architecture must instead select authoritative current versions and surface unresolved conflicts with citations; a separate change is still needed for version-aware authoritative evidence.

Question 178

A query requires a complete list of all overdue invoices, but top-k vector retrieval only finds a sample. Which architecture adjustment is most appropriate?

  1. Use a paginated database query with completeness checks.
  2. Use semantic vector retrieval with an evaluated embedding model.
  3. Use hybrid lexical-semantic retrieval with reranking tested on both query families.
  4. Use a controlled SQL or analytics query and verify the computation before natural-language explanation.

Explanation

Correct Answer

A:

Top-k similarity is a relevance technique, not an exhaustive records enumeration service. Build query cases that verify complete entity-set retrieval and check whether the chosen method returns complete, applicable evidence rather than merely similar text.

Incorrect Answers

B:

This recommendation targets paraphrase-aware document retrieval at a different point in the workflow. It leaves complete entity-set retrieval outside the design contract and does not require the team to use a paginated database query with completeness checks.

C:

This is a valid architectural concern when optimizing mixed identifier-and-semantic search. Here it leaves complete entity-set retrieval unaddressed, whereas the intended change must use a paginated database query with completeness checks before the case can be considered resolved.

D:

This would make sense as an additional project activity for exact numerical aggregation from structured data. The priority required here is complete entity-set retrieval, so the workflow still needs a control that will use a paginated database query with completeness checks.

Question 179

A Claude application must discover standardized tool definitions from several external services. Which next step is most defensible?

  1. Use MCP servers with narrowly scoped tools and authenticated transports.
  2. Use a defined agent-to-agent contract with identity, task state and handoff validation.
  3. Choose a supported network transport with authenticated endpoints and server-side authorization.
  4. Wrap API or MCP mutation paths in transactional logging and idempotency controls.

Explanation

Correct Answer

A:

MCP offers a common tool and resource interface; authorization still belongs to service and gateway controls. Validate standardized discovery across tool services at the execution host, including transport reachability and identity; protocol compatibility does not authorize standardized discovery across tool services.

Incorrect Answers

B:

This shifts attention toward authenticated inter-agent handoffs when the relevant decision is about standardized discovery across tool services. Unless the architecture can use MCP servers with narrowly scoped tools and authenticated transports, the failure described in the stem remains unresolved.

C:

For a workflow whose primary weakness was deployable remote-tool connectivity, that proposal could be reasonable. Here the unmet obligation is standardized discovery across tool services, and that requires the system to use MCP servers with narrowly scoped tools and authenticated transports.

D:

Although transactional tool operation governance can be valuable, it is not the acceptance condition in the question. The missing provision is standardized discovery across tool services, addressed by the control to use MCP servers with narrowly scoped tools and authenticated transports.

Question 180

A backend needs one high-volume, deterministic inventory lookup with a stable schema. Which approach should be recommended?

  1. Implement the application-side tool execution and return the resulting tool message.
  2. Retain the existing API adapter unless MCP adds a demonstrated interoperability benefit.
  3. Use a defined agent-to-agent contract with identity, task state and handoff validation.
  4. Call the inventory API directly from an application service.

Explanation

Correct Answer

D:

A direct API avoids unnecessary protocol and model-discovery overhead for a fixed operation. Validate efficient fixed API integration at the execution host, including transport reachability and identity; protocol compatibility does not authorize efficient fixed api integration.

Incorrect Answers

A:

That choice would be responsive to correct client tool execution loop, but the reported defect is different. It leaves efficient fixed api integration unproved because there is no commitment to call the inventory API directly from an application service.

B:

There is a legitimate use case for avoiding unneeded connector replatforming, but it does not supply the test or enforcement needed for efficient fixed api integration. The design needs to call the inventory API directly from an application service.

C:

Consider its actual operating scope: authenticated inter-agent handoffs. The requested assurance is efficient fixed api integration, which calls for an implementation that can call the inventory API directly from an application service, not merely for a related improvement.

Claude Certified Architect - Professional (CCAR-P) Practice Tests