Anthropic CCDV-F (Claude Certified Developer - Foundations) Practice Test 65

Question 640

Engineers are splitting a tool integration service into two independently tested functions. Function A must use an already supported capability without hosting a duplicate custom service, while function B must enable a customer service agent to prepare but not autonomously approve account closures. Which plan is sound?

  1. First component: Use system or repository instructions for stable behavior. Second component: Combine tool access with human confirmation.
  2. First component: Choose a built-in provider tool. Second component: Combine tool access with human confirmation.
  3. First component: Choose a built-in provider tool. Second component: Create a custom client-side tool.
  4. First component: Create a custom client-side tool. Second component: Use system or repository instructions for stable behavior.

Explanation

Correct Answer

B:

The first proposal fits its assigned role because provider-managed execution can reduce implementation work when its scope matches the requirement, while the second proposal fits its assigned role because tools supply actionable capabilities while a separate approval step governs sensitive execution; their respective acceptance criteria are to use an already supported capability without hosting a duplicate custom service and to enable a customer service agent to prepare but not autonomously approve account closures.

Incorrect Answers

A:

The first proposal serves the different purpose to make a coding or API assistant follow organization-wide conventions instead of the assigned first role because instructions shape model behavior and task framing without executing external operations, while the second proposal fits its assigned role because tools supply actionable capabilities while a separate approval step governs sensitive execution; their respective acceptance criteria are to use an already supported capability without hosting a duplicate custom service and to enable a customer service agent to prepare but not autonomously approve account closures.

C:

The first proposal fits its assigned role because provider-managed execution can reduce implementation work when its scope matches the requirement, while the second proposal serves the different purpose to call a protected internal pricing API under the application's own credentials instead of the assigned second role because local execution gives the developer direct control over authentication, validation, and side effects; their respective acceptance criteria are to use an already supported capability without hosting a duplicate custom service and to enable a customer service agent to prepare but not autonomously approve account closures.

D:

The first proposal serves the different purpose to call a protected internal pricing API under the application's own credentials instead of the assigned first role because local execution gives the developer direct control over authentication, validation, and side effects, while the second proposal serves the different purpose to make a coding or API assistant follow organization-wide conventions instead of the assigned second role because instructions shape model behavior and task framing without executing external operations; their respective acceptance criteria are to use an already supported capability without hosting a duplicate custom service and to enable a customer service agent to prepare but not autonomously approve account closures.

Question 641

The owner of a prompt engineering integration wants to prevent a customer message from overriding the application's escalation rules. Which proposal provides the required mechanism?

  1. Separate system policy from user content
  2. State one measurable outcome per instruction
  3. Use few-shot boundary cases
  4. Specify output format in the prompt

Explanation

Correct Answer

A:

Stable higher-priority instructions should be represented in the designated instruction channel; this behavior directly supports the need to prevent a customer message from overriding the application's escalation rules. User text is the task data, not a privileged configuration surface.

Incorrect Answers

B:

Clear task goals and acceptance criteria reduce competing interpretations; despite being useful when a service must stop a summarizer from producing unrelated analysis instead of required action items, it is not a solution to prevent a customer message from overriding the application's escalation rules; a vague request for a good summary provides no testable target.

C:

For the alternate objective to distinguish near-duplicate support intents that a broad rule routinely confuses, examples illustrating each side of a decision boundary communicate nuanced classification criteria; the feature here must prevent a customer message from overriding the application's escalation rules, and many unrelated demonstrations can make the rule harder to infer.

D:

Explicit field descriptions guide generation toward the consumer's expected structure; this applies to the different goal to obtain named fields in a consistent response layout for subsequent processing, not the requirement to prevent a customer message from overriding the application's escalation rules; formatting guidance still requires machine validation at the boundary.

Question 642

The current design of a language-model integration pipeline does not adequately avoid selecting a model that lacks a required document or tool feature. Which improvement resolves the gap?

  1. Prefer a Haiku-class model for lightweight routing
  2. Check modality and feature support
  3. Prefer a Sonnet-class model for balanced workloads
  4. Measure end-to-end latency, not just token speed

Explanation

Correct Answer

B:

Availability of context length, structured output, and modalities may depend on the selected model and version; this behavior directly supports the need to avoid selecting a model that lacks a required document or tool feature. Family-level generalizations do not override specific compatibility documentation.

Incorrect Answers

A:

The lighter tier suits inexpensive, high-throughput tasks with modest reasoning demand; this applies to the different goal to classify many short messages where low latency and unit cost dominate, not the requirement to avoid selecting a model that lacks a required document or tool feature; complex planning may need escalation after an evaluated failure.

C:

The balanced tier often offers a practical quality-throughput compromise, subject to task-level testing; the supported outcome is to serve a mixed coding and support assistant at moderate latency and cost, whereas this scenario requires the ability to avoid selecting a model that lacks a required document or tool feature; it should not be assumed to dominate every model on every benchmark.

D:

This approach serves the separate need to select the right model for an application with tool and network overhead because total response time includes queuing, input processing, generation, and downstream integration; it does not establish the outcome to avoid selecting a model that lacks a required document or tool feature, and a high tokens-per-second figure alone can conceal slow first-token delivery.

Question 643

A document processing review requires two independent, explicitly defined outcomes: the first workstream must detect unsafe permissions and schema changes before merging an agent tool; a second must serve independent API calls without blocking a worker thread for each network wait. Which TWO actions directly provide these capabilities? Choose TWO.

  1. Apply dependency injection for external clients
  2. Separate integration tests from unit tests
  3. Use feature flags for risky behavior changes
  4. Use asynchronous I/O for concurrent requests
  5. Use pull requests and focused code review

Explanation

Correct Answers

D:

Async execution overlaps I/O-bound tasks while maintaining control over concurrency; this directly supports task 2 to serve independent API calls without blocking a worker thread for each network wait, not the separate task or tasks to detect unsafe permissions and schema changes before merging an agent tool; it does not by itself make CPU-bound model processing faster.

E:

Peer review creates an explicit checkpoint for correctness, security, and maintainability; this directly supports task 1 to detect unsafe permissions and schema changes before merging an agent tool, not the separate task or tasks to serve independent API calls without blocking a worker thread for each network wait; running a linter alone cannot assess authorization intent.

Incorrect Answers

A:

Injected interfaces allow deterministic stubs and reduce coupling to vendors; this implements the different outcome to test application logic without issuing real model calls or payments rather than the required tasks to serve independent API calls without blocking a worker thread for each network wait and to detect unsafe permissions and schema changes before merging an agent tool, and hard-coded global clients make behavior harder to isolate and simulate, so it is not a direct solution to any of the assigned workstreams.

B:

Integration tests exercise boundaries and serialization that isolated mocks may miss; this implements the different outcome to verify that a real staging tool and its API contract work together rather than the required tasks to serve independent API calls without blocking a worker thread for each network wait and to detect unsafe permissions and schema changes before merging an agent tool, and unit tests can pass while authentication or payload mapping is broken, so it is not a direct solution to any of the assigned workstreams.

C:

Runtime controls can segment rollout and permit quick disabling without a code rebuild; this implements the different outcome to activate a new tool execution path for only internal testers rather than the required tasks to serve independent API calls without blocking a worker thread for each network wait and to detect unsafe permissions and schema changes before merging an agent tool, and hard-coded unconditional activation exposes every tenant simultaneously, so it is not a direct solution to any of the assigned workstreams.

Question 644

A production API application is approaching release, but the team still needs to trace a quality regression to the exact instruction revision used. What is the most appropriate implementation?

  1. Store API keys in managed secret storage
  2. Put repository guidance in CLAUDE.md
  3. Track MCP server versions and contracts
  4. Version prompt templates alongside code

Explanation

Correct Answer

D:

Source-controlled prompt assets make rollbacks, diffs, and evaluations possible; this behavior directly supports the need to trace a quality regression to the exact instruction revision used. Editing prompts in an unversioned dashboard hides causal history.

Incorrect Answers

A:

When an application needs to rotate production credentials without committing new source code, central secret management supports scoped access, auditing, and key rotation; the current capability must instead trace a quality regression to the exact instruction revision used, while a plain-text config checked into a repository is an inappropriate credential store.

B:

Repository memory files provide persistent instructional context to coding sessions; the supported outcome is to give Claude Code the project's build, test, and architecture conventions, whereas this scenario requires the ability to trace a quality regression to the exact instruction revision used; they are not an application secrets manager or the API's global system prompt.

C:

This approach serves the separate need to keep external tool connectors compatible after client upgrades because declared versions and integration tests expose resource or schema changes before deployment; it does not establish the outcome to trace a quality regression to the exact instruction revision used, and automatically using the newest connector can alter behavior unexpectedly.

Question 645

Before an incident response release, three separately owned tasks must be covered: one service must answer routine account questions without the overhead of orchestrating multiple agents; another must ensure a generated specification is validated before code generation begins; and a third must coordinate separate billing, compliance, and engineering analyses under one final decision. Which THREE choices directly address the stated tasks? Choose THREE.

  1. Use an agentic decision loop
  2. Pass structured handoff artifacts between workers
  3. Keep a single agent with a small tool set
  4. Use parallel fan-out and a reconciliation step
  5. Adopt a supervisor with specialist workers
  6. Use sequential chaining for dependent subtasks

Explanation

Correct Answers

C:

A single constrained executor simplifies state, authentication, logging, and latency where specialization adds little; this directly supports task 1 to answer routine account questions without the overhead of orchestrating multiple agents, not the separate task or tasks to ensure a generated specification is validated before code generation begins and to coordinate separate billing, compliance, and engineering analyses under one final decision; a multi-agent hierarchy is not a default requirement for tool use.

E:

A managing agent delegates bounded subtasks to specialists and reconciles their outputs against a common objective; this directly supports task 3 to coordinate separate billing, compliance, and engineering analyses under one final decision, not the separate task or tasks to answer routine account questions without the overhead of orchestrating multiple agents and to ensure a generated specification is validated before code generation begins; independent workers with no coordinator can return inconsistent recommendations.

F:

Each stage consumes an approved artifact from the previous stage, preserving dependency order and stage-specific checks; this directly supports task 2 to ensure a generated specification is validated before code generation begins, not the separate task or tasks to answer routine account questions without the overhead of orchestrating multiple agents and to coordinate separate billing, compliance, and engineering analyses under one final decision; parallel execution would allow downstream work to proceed on unverified assumptions.

Incorrect Answers

A:

A model-directed loop can choose subsequent actions based on tool observations until it reaches a stopping condition; this implements the different outcome to investigate an open-ended incident where the next query depends on earlier findings rather than the required tasks to coordinate separate billing, compliance, and engineering analyses under one final decision and to answer routine account questions without the overhead of orchestrating multiple agents and to ensure a generated specification is validated before code generation begins, and a predetermined chain cannot reliably anticipate all branches of an exploratory diagnosis, so it is not a direct solution to any of the assigned workstreams.

B:

Defined input and output schemas make it possible to attribute conclusions and detect missing fields at the handoff; this implements the different outcome to preserve evidence and data lineage when one agent delegates a subtask to another rather than the required tasks to coordinate separate billing, compliance, and engineering analyses under one final decision and to answer routine account questions without the overhead of orchestrating multiple agents and to ensure a generated specification is validated before code generation begins, and free-form chat summaries can omit identifiers required by the next worker, so it is not a direct solution to any of the assigned workstreams.

D:

Parallel tasks reduce critical-path latency when they do not depend on one another; a join step resolves conflicting findings; this implements the different outcome to analyze several independent document sections concurrently and combine the results rather than the required tasks to coordinate separate billing, compliance, and engineering analyses under one final decision and to answer routine account questions without the overhead of orchestrating multiple agents and to ensure a generated specification is validated before code generation begins, and a serial chain makes independent stages wait without improving correctness, so it is not a direct solution to any of the assigned workstreams.

Question 646

A tool integration review requires two independent, explicitly defined outcomes: the first workstream must share a repeatable analyst workflow across supported clients; a second must verify that tools and resources are discoverable before integrating an agent. Which TWO actions directly provide these capabilities? Choose TWO.

  1. Expose an MCP prompt template
  2. Test protocol behavior with a compliant inspector
  3. Validate MCP tool schemas and results
  4. Expose an MCP tool for an external action
  5. Expose an MCP resource for reference content

Explanation

Correct Answers

A:

Prompts describe reusable message templates or workflows offered by the server; this directly supports task 1 to share a repeatable analyst workflow across supported clients, not the separate task or tasks to verify that tools and resources are discoverable before integrating an agent; a prompt template does not itself execute a database update.

B:

A protocol test client exposes handshake, capability, and schema problems independently of model behavior; this directly supports task 2 to verify that tools and resources are discoverable before integrating an agent, not the separate task or tasks to share a repeatable analyst workflow across supported clients; debugging the assistant's prompt is ineffective when the server cannot negotiate correctly.

Incorrect Answers

C:

Contract tests check declared capabilities, argument shapes, and expected response content; this implements the different outcome to prevent a server update from returning incompatible payloads to clients rather than the required tasks to share a repeatable analyst workflow across supported clients and to verify that tools and resources are discoverable before integrating an agent, and successful connection does not prove semantic correctness, so it is not a direct solution to any of the assigned workstreams.

D:

MCP tools advertise callable operations with named parameters and schemas across client integrations; this implements the different outcome to let multiple compatible clients request a controlled shipment lookup rather than the required tasks to share a repeatable analyst workflow across supported clients and to verify that tools and resources are discoverable before integrating an agent, and a resource intended only for reading context does not perform arbitrary commands, so it is not a direct solution to any of the assigned workstreams.

E:

Resources provide identifiable data that clients can fetch for model context; this implements the different outcome to make a versioned product catalog available for read-oriented context retrieval rather than the required tasks to share a repeatable analyst workflow across supported clients and to verify that tools and resources are discoverable before integrating an agent, and treating static content as a privileged executable tool unnecessarily expands authority, so it is not a direct solution to any of the assigned workstreams.

Question 647

A prompt engineering solution has parallel requirements with separate owners. Workstream one must prevent a negative invoice total from reaching a payment ledger; workstream two must detect plausible dates or amounts that are absent from the provided document. Which combination meets both?

  1. First component: Validate required schema fields. Second component: Check evidence alignment after extraction.
  2. First component: Validate semantic business invariants. Second component: Check evidence alignment after extraction.
  3. First component: Validate semantic business invariants. Second component: Parse JSON with a real parser.
  4. First component: Parse JSON with a real parser. Second component: Validate required schema fields.

Explanation

Correct Answer

B:

The first proposal fits its assigned role because domain rules check value ranges and cross-field relationships beyond syntactic shape, while the second proposal fits its assigned role because semantic verification compares fields to authoritative source spans; their respective acceptance criteria are to prevent a negative invoice total from reaching a payment ledger and to detect plausible dates or amounts that are absent from the provided document.

Incorrect Answers

A:

The first proposal serves the different purpose to reject a response that omits a mandated account identifier instead of the assigned first role because schema checks enforce field presence, types, and enumerations before the result enters a database, while the second proposal fits its assigned role because semantic verification compares fields to authoritative source spans; their respective acceptance criteria are to prevent a negative invoice total from reaching a payment ledger and to detect plausible dates or amounts that are absent from the provided document.

C:

The first proposal fits its assigned role because domain rules check value ranges and cross-field relationships beyond syntactic shape, while the second proposal serves the different purpose to avoid accepting malformed generated objects because they resemble JSON to a reader instead of the assigned second role because syntax validation converts serialized text into a trustworthy data structure only after parsing succeeds; their respective acceptance criteria are to prevent a negative invoice total from reaching a payment ledger and to detect plausible dates or amounts that are absent from the provided document.

D:

The first proposal serves the different purpose to avoid accepting malformed generated objects because they resemble JSON to a reader instead of the assigned first role because syntax validation converts serialized text into a trustworthy data structure only after parsing succeeds, while the second proposal serves the different purpose to reject a response that omits a mandated account identifier instead of the assigned second role because schema checks enforce field presence, types, and enumerations before the result enters a database; their respective acceptance criteria are to prevent a negative invoice total from reaching a payment ledger and to detect plausible dates or amounts that are absent from the provided document.

Question 648

A workflow automation group is preparing a production change intended to enforce approval, execution, and confirmation order in a compliance workflow. Which technique matches that intent?

  1. Prefer a finite state machine for regulated steps
  2. Use an independent verifier step
  3. Use a framework only when its primitives fit
  4. Use event logs for tool actions

Explanation

Correct Answer

A:

Explicit states and transitions control who may advance a case independently of generated language; this behavior directly supports the need to enforce approval, execution, and confirmation order in a compliance workflow. An unrestricted agent loop can skip expected governance stages.

Incorrect Answers

B:

A separate validation stage checks the artifact against explicit tests and constraints, limiting self-confirming errors; the supported outcome is to detect mistakes in an agent-generated SQL migration before running it, whereas this scenario requires the ability to enforce approval, execution, and confirmation order in a compliance workflow; asking the original agent whether it is correct is weaker than executable verification.

C:

This approach serves the separate need to select orchestration dependencies for a project that already has a stable task engine because frameworks can help with graphs, retries, and checkpoints but create maintenance and integration costs; it does not establish the outcome to enforce approval, execution, and confirmation order in a compliance workflow, and adopting a framework solely for fashionable abstraction is not a technical requirement.

D:

Structured traces connect tool intent, arguments, results, and subsequent decisions for audit and replay; despite being useful when a service must explain why an agent modified a configuration during a failed automated release, it is not a solution to enforce approval, execution, and confirmation order in a compliance workflow; a final chat summary omits the exact causal sequence.

Question 649

A developer is reviewing alternatives for an engineering productivity feature that must prevent a slow external inventory service from blocking an agent indefinitely. Which option is a defensible fit?

  1. Forward tool_result blocks with matching identifiers
  2. Implement a custom Messages API tool loop
  3. Keep user-specific state out of a shared agent process
  4. Set a tool execution timeout

Explanation

Correct Answer

D:

Deadlines bound latency and convert stalled calls into explicit recoverable failures; this behavior directly supports the need to prevent a slow external inventory service from blocking an agent indefinitely. A generous model output limit cannot constrain external network operations.

Incorrect Answers

A:

For the alternate objective to let Claude continue after a client tool returns a transaction status, results must reference the corresponding tool_use IDs and be included in the next conversation turn; the feature here must prevent a slow external inventory service from blocking an agent indefinitely, and sending a human-readable summary without the tool correlation can break the protocol.

B:

The application receives tool requests, validates inputs, executes authorized functions, and sends tool results back; this applies to the different goal to retain complete control over a small set of proprietary tool calls and their execution order, not the requirement to prevent a slow external inventory service from blocking an agent indefinitely; the Messages API does not automatically execute arbitrary client tools.

C:

When an application needs to serve many tenants without mixing their tool outputs or prior messages, per-request identity and session boundaries prevent cross-tenant context leakage; the current capability must instead prevent a slow external inventory service from blocking an agent indefinitely, while shared conversation histories can inadvertently expose unrelated customers' data.

Question 650

During a document processing rollout, two different failure paths need remediation. The first component must decide which customer fields must be redacted before model requests; separately, the second must deliver model-generated records to a strict financial reporting API. Which changes should be adopted?

  1. First component: Classify data sensitivity up front. Second component: Identify authoritative source systems.
  2. First component: Identify authoritative source systems. Second component: Define failure and recovery objectives.
  3. First component: Classify data sensitivity up front. Second component: Agree on output contracts with downstream systems.
  4. First component: Define failure and recovery objectives. Second component: Agree on output contracts with downstream systems.

Explanation

Correct Answer

C:

The first proposal fits its assigned role because data classification guides minimization, access controls, retention, and acceptable transfers, while the second proposal fits its assigned role because typed fields and required identifiers enable validators and consumers to reject malformed results; their respective acceptance criteria are to decide which customer fields must be redacted before model requests and to deliver model-generated records to a strict financial reporting API.

Incorrect Answers

A:

The first proposal fits its assigned role because data classification guides minimization, access controls, retention, and acceptable transfers, while the second proposal serves the different purpose to prevent a Claude assistant from inventing account balances when it lacks current ledger data instead of the assigned second role because source-of-truth mapping identifies which verified systems must supply facts and what freshness guarantees apply; their respective acceptance criteria are to decide which customer fields must be redacted before model requests and to deliver model-generated records to a strict financial reporting API.

B:

The first proposal serves the different purpose to prevent a Claude assistant from inventing account balances when it lacks current ledger data instead of the assigned first role because source-of-truth mapping identifies which verified systems must supply facts and what freshness guarantees apply, while the second proposal serves the different purpose to keep a regulated triage service usable during an upstream API outage instead of the assigned second role because availability, recovery time, and fallback behavior must be specified as measurable operational needs; their respective acceptance criteria are to decide which customer fields must be redacted before model requests and to deliver model-generated records to a strict financial reporting API.

D:

The first proposal serves the different purpose to keep a regulated triage service usable during an upstream API outage instead of the assigned first role because availability, recovery time, and fallback behavior must be specified as measurable operational needs, while the second proposal fits its assigned role because typed fields and required identifiers enable validators and consumers to reject malformed results; their respective acceptance criteria are to decide which customer fields must be redacted before model requests and to deliver model-generated records to a strict financial reporting API.

Claude Certified Developer - Foundations (CCDV-F) Practice Tests