Anthropic CCDV-F (Claude Certified Developer - Foundations) Practice Test 26

Question 250

A hypothetical generator succeeds on each independent attempt with probability 0.8. A perfect validator recognizes success, and the system tries at most twice. What is the probability of at least one success?

  1. 1.60.
  2. 0.96.
  3. 1.00.
  4. 0.80.

Explanation

Correct Answer

B:

Both attempts fail with probability 0.2 × 0.2 = 0.04. Subtracting that joint failure from one gives 0.96; the calculation depends on the stated independence and perfect-validator assumptions.

Incorrect Answers

A:

Adding two success probabilities double-counts outcomes where both succeed and exceeds the valid range for a probability.

C:

Two attempts improve the chance of success but do not eliminate the 0.04 probability that both independently fail.

D:

This is the success probability of one attempt. It omits the chance that the second attempt succeeds after the first fails.

Question 251

Why can retrieval improve answers about a frequently updated internal policy?

  1. Retrieval removes the need to authorize documents.
  2. Retrieval permanently retrains all future conversations.
  3. Retrieval guarantees every conclusion is correct.
  4. It supplies current task-specific evidence beyond the model’s static learned knowledge.

Explanation

Correct Answer

D:

Retrieval supplies current policy evidence at inference time. The model can use that evidence without a weight update for every document revision.

Incorrect Answers

A:

Access control remains necessary before supplying private evidence.

B:

Providing evidence at inference time does not automatically update shared weights.

C:

Relevant input helps, but reasoning and output validation still matter.

Question 252

A job endpoint returns HTTP 202 Accepted. What should the caller generally distinguish?

  1. Acceptance from authentication failure.
  2. Acceptance for processing from completed business execution.
  3. Acceptance from proof that no further status check is needed.
  4. Acceptance from a guaranteed transaction rollback.

Explanation

Correct Answer

B:

HTTP 202 indicates acceptance for processing. Follow the endpoint's job-status or completion-event mechanism before reporting completed business execution.

Incorrect Answers

A:

A 202 response does not itself report invalid credentials.

C:

Asynchronous processing can require observing a later outcome.

D:

The status does not establish rollback semantics.

Question 253

An endpoint expects JSON, but receives a form-encoded body with a JSON content-type declaration. What is wrong?

  1. The endpoint must ignore the declared type and infer any body encoding.
  2. The actual encoding and declared representation disagree.
  3. Changing only the content-type header automatically converts form encoding into JSON.
  4. Matching form field names make the body equivalent to a JSON object.

Explanation

Correct Answer

B:

The declared media type must match the body encoding. A JSON declaration attached to form-encoded bytes can cause the server to apply an unsuitable parser.

Incorrect Answers

A:

The endpoint has a documented JSON interface. Depending on automatic format guessing does not repair the client's mismatch with that contract.

C:

Headers describe the representation; they do not rewrite the body into a different serialization format.

D:

Matching names do not make two wire encodings interchangeable. The server must receive bytes encoded according to its declared JSON request contract.

Question 254

HTTPS encrypts transport. Does that alone authorize a caller to read a private customer record?

  1. Yes; encryption grants every connected user identical privileges.
  2. Yes; certificates replace all application access checks.
  3. No; HTTPS provides no transport protection.
  4. No; authentication and resource authorization remain separate requirements.

Explanation

Correct Answer

D:

Encryption protects the communication channel, while application authorization separately checks which records the authenticated caller may read.

Incorrect Answers

A:

Secure transport is not an entitlement system.

B:

Server identity validation does not establish every user’s resource rights.

C:

That denies its real purpose rather than explaining its limits.

Question 255

A connection succeeds quickly, but the remote service stops sending data. Which timeout distinction matters?

  1. Only the total number of generated tokens allowed.
  2. Only the deadline for the initial DNS lookup.
  3. Only the connection-establishment timeout.
  4. Connection establishment versus waiting for response data.

Explanation

Correct Answer

D:

A successful connection does not bound subsequent reads; the client needs appropriate response or overall deadlines.

Incorrect Answers

A:

Output capacity does not set how long a stalled network read may wait without receiving data.

B:

The connection was already established, so limiting name resolution does not address the subsequent stall in response data.

C:

The connection already succeeded; the observed stall occurs while waiting for subsequent response data.

Question 256

An application needs continuous independent messages in both directions over one established channel. Which transport capability is relevant?

  1. A bidirectional channel such as WebSocket.
  2. A single completed HTTP response that closes after returning one result.
  3. SSE alone as a symmetric two-way protocol.
  4. An SSE stream alone used as the only channel in both directions.

Explanation

Correct Answer

A:

This fits ongoing client-to-server and server-to-client communication; SSE primarily delivers server events to clients.

Incorrect Answers

B:

A closed request-response exchange does not supply an ongoing bidirectional channel for independent later messages.

C:

SSE’s event stream is server-to-client; separate requests are needed for client messages.

D:

SSE is a server-to-client event stream; independent client messages require a separate request path or another bidirectional mechanism.

Question 257

Profiling shows many short API calls spend substantial time establishing fresh TCP and TLS connections. What client change can reduce this overhead when supported?

  1. Reuse connections through a bounded connection pool.
  2. Put previous responses into the next request body.
  3. Increase the maximum output tokens for each request.
  4. Create a new HTTP client and connection for every request.

Explanation

Correct Answer

A:

An existing connection can serve another compatible request without repeating the full establishment sequence. Pool limits and idle-connection handling remain necessary to manage resources and failures.

Incorrect Answers

B:

Conversation context does not reuse transport connections. It can increase input size while leaving handshake overhead unchanged.

C:

Generation capacity does not remove TCP or TLS setup work. It addresses a different part of request execution.

D:

This preserves the repeated setup cost identified by the profile and prevents effective connection reuse across calls.

Question 258

The API streams promptly, but users see all text at once behind a reverse proxy. What should be checked?

  1. The assumption that stream=true controls every network intermediary.
  2. Proxy buffering and end-to-end streaming configuration.
  3. Only whether the client sends a valid JSON request body.
  4. Only the model's total answer length after the proxy finishes.

Explanation

Correct Answer

B:

A buffering proxy can collect incremental chunks before forwarding them, hiding streaming benefits despite timely events from the upstream API.

Incorrect Answers

A:

Upstream configuration does not automatically configure the complete delivery path.

C:

A valid request already reached the API; the delayed delivery of incremental output points to the response transport path.

D:

Final length does not show whether the intermediary buffered chunks instead of forwarding them as they arrived.

Question 259

Two implementations finish in the same total time, but one displays useful text much earlier. Which metric captures that difference?

  1. Time to first useful output or first token.
  2. Only the number of input tokens in each request.
  3. Only total output-token count.
  4. Only the time at which the final response completes.

Explanation

Correct Answer

A:

Time to first useful output captures perceived responsiveness, while total completion time measures when the entire response is available.

Incorrect Answers

B:

Input size can affect latency but does not directly measure when the user first receives useful text.

C:

Equal or different token counts do not directly establish when display begins.

D:

Both implementations have equal total duration, so that metric misses the difference in early usable output.

Question 260

A network read returns half a protocol event. What must the parser do?

  1. Discard every incomplete chunk immediately.
  2. Retry the whole generation after each small read.
  3. Treat every read as one complete JSON object.
  4. Buffer until a complete event boundary is available.

Explanation

Correct Answer

D:

Transport chunks need not align with application messages; parsing must follow protocol framing rather than read-call boundaries.

Incorrect Answers

A:

Discarding fragments loses data required to reconstruct the event.

B:

Small chunks are normal streaming behavior, not necessarily failures.

C:

A read can split or combine messages, making that assumption unsafe.

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