Anthropic CCDV-F (Claude Certified Developer - Foundations) Practice Test 1

Question 1

Two features of an agent platform system are under review. One feature must reduce turnaround when an agent asks for unrelated reads in the same response; the other must recover agent state when a container restarts during a lengthy customer case. Which plan matches the respective needs?

  1. First component: Execute independent requested tools concurrently. Second component: Choose isolated sandbox execution for untrusted code.
  2. First component: Choose isolated sandbox execution for untrusted code. Second component: Use the Claude Agent SDK for an integrated tool-driven agent.
  3. First component: Use the Claude Agent SDK for an integrated tool-driven agent. Second component: Use a durable session store.
  4. First component: Execute independent requested tools concurrently. Second component: Use a durable session store.

Explanation

Correct Answer

D:

The first proposal fits its assigned role because parallel execution is safe for independent operations while every result must still be matched to its call ID, while the second proposal fits its assigned role because external persistence holds the conversation and workflow metadata needed to resume safely; their respective acceptance criteria are to reduce turnaround when an agent asks for unrelated reads in the same response and to recover agent state when a container restarts during a lengthy customer case.

Incorrect Answers

A:

The first proposal fits its assigned role because parallel execution is safe for independent operations while every result must still be matched to its call ID, while the second proposal serves the different purpose to let an agent run generated scripts without exposing production host files instead of the assigned second role because a restricted execution environment limits filesystem, network, and resource privileges; their respective acceptance criteria are to reduce turnaround when an agent asks for unrelated reads in the same response and to recover agent state when a container restarts during a lengthy customer case.

B:

The first proposal serves the different purpose to let an agent run generated scripts without exposing production host files instead of the assigned first role because a restricted execution environment limits filesystem, network, and resource privileges, while the second proposal serves the different purpose to build a coding agent that needs managed tool execution, sessions, and event handling instead of the assigned second role because the SDK provides agent-loop primitives and integrations that remove much of the custom harness work; their respective acceptance criteria are to reduce turnaround when an agent asks for unrelated reads in the same response and to recover agent state when a container restarts during a lengthy customer case.

C:

The first proposal serves the different purpose to build a coding agent that needs managed tool execution, sessions, and event handling instead of the assigned first role because the SDK provides agent-loop primitives and integrations that remove much of the custom harness work, while the second proposal fits its assigned role because external persistence holds the conversation and workflow metadata needed to resume safely; their respective acceptance criteria are to reduce turnaround when an agent asks for unrelated reads in the same response and to recover agent state when a container restarts during a lengthy customer case.

Question 2

A data pipeline build contains an upstream function and a separate downstream function. The upstream work must prevent a malicious help article from ordering the model to reveal secrets; the downstream work must prevent a new user from inheriting another user's document excerpts. Which configuration pair covers both?

  1. First component: Treat retrieval results as untrusted data. Second component: Limit retained context to authorized history.
  2. First component: Separate model reasoning from authoritative calculations. Second component: Use bounded retrieval results.
  3. First component: Treat retrieval results as untrusted data. Second component: Separate model reasoning from authoritative calculations.
  4. First component: Handle empty and partial model responses. Second component: Separate model reasoning from authoritative calculations.
  5. First component: Use bounded retrieval results. Second component: Limit retained context to authorized history.

Explanation

Correct Answer

A:

The first proposal fits its assigned role because retrieved content should be framed as evidence, never upgraded to new tool or system authority, while the second proposal fits its assigned role because context assembly must apply access checks before adding prior turns or retrieved evidence; their respective acceptance criteria are to prevent a malicious help article from ordering the model to reveal secrets and to prevent a new user from inheriting another user's document excerpts.

Incorrect Answers

B:

The first proposal serves the different purpose to compute discounts through verified business code rather than generated arithmetic alone instead of the assigned first role because deterministic services can enforce exact formulas and rounding before the model explains the result, while the second proposal serves the different purpose to avoid burying the relevant contract clause beneath hundreds of low-value passages instead of the assigned second role because ranking and a context budget prioritize high-quality evidence while controlling input tokens; their respective acceptance criteria are to prevent a malicious help article from ordering the model to reveal secrets and to prevent a new user from inheriting another user's document excerpts.

C:

The first proposal fits its assigned role because retrieved content should be framed as evidence, never upgraded to new tool or system authority, while the second proposal serves the different purpose to compute discounts through verified business code rather than generated arithmetic alone instead of the assigned second role because deterministic services can enforce exact formulas and rounding before the model explains the result; their respective acceptance criteria are to prevent a malicious help article from ordering the model to reveal secrets and to prevent a new user from inheriting another user's document excerpts.

D:

The first proposal serves the different purpose to keep a downstream parser from crashing when a request returns no usable text block instead of the assigned first role because content-aware response handling checks block types, stop reasons, and required values before proceeding, while the second proposal serves the different purpose to compute discounts through verified business code rather than generated arithmetic alone instead of the assigned second role because deterministic services can enforce exact formulas and rounding before the model explains the result; their respective acceptance criteria are to prevent a malicious help article from ordering the model to reveal secrets and to prevent a new user from inheriting another user's document excerpts.

E:

The first proposal serves the different purpose to avoid burying the relevant contract clause beneath hundreds of low-value passages instead of the assigned first role because ranking and a context budget prioritize high-quality evidence while controlling input tokens, while the second proposal fits its assigned role because context assembly must apply access checks before adding prior turns or retrieved evidence; their respective acceptance criteria are to prevent a malicious help article from ordering the model to reveal secrets and to prevent a new user from inheriting another user's document excerpts.

Question 3

A language-model integration build contains an upstream function and a separate downstream function. The upstream work must reduce setup overhead when a service sends many short requests; the downstream work must simplify authentication, serialization, and response typing in a Python service. Which configuration pair covers both?

  1. First component: Use connection pooling for repeated HTTP calls. Second component: Use direct HTTP REST integration for unsupported runtimes.
  2. First component: Choose WebSockets for bidirectional realtime exchange. Second component: Use direct HTTP REST integration for unsupported runtimes.
  3. First component: Use direct HTTP REST integration for unsupported runtimes. Second component: Use a queue for burst absorption.
  4. First component: Use a queue for burst absorption. Second component: Use the official SDK for standard requests.
  5. First component: Use connection pooling for repeated HTTP calls. Second component: Use the official SDK for standard requests.

Explanation

Correct Answer

E:

The first proposal fits its assigned role because reusable clients avoid unnecessary connection establishment under compatible networking conditions, while the second proposal fits its assigned role because maintained client libraries wrap the HTTP API and expose structured response objects; their respective acceptance criteria are to reduce setup overhead when a service sends many short requests and to simplify authentication, serialization, and response typing in a Python service.

Incorrect Answers

A:

The first proposal fits its assigned role because reusable clients avoid unnecessary connection establishment under compatible networking conditions, while the second proposal serves the different purpose to call the Messages endpoint from a language without a mature provider SDK instead of the assigned second role because a standards-based HTTP client can send JSON requests with the correct headers and decode responses; their respective acceptance criteria are to reduce setup overhead when a service sends many short requests and to simplify authentication, serialization, and response typing in a Python service.

B:

The first proposal serves the different purpose to implement an interactive collaboration channel requiring messages in both directions instead of the assigned first role because persistent two-way channels fit interactive event exchange where the client also sends messages during a session, while the second proposal serves the different purpose to call the Messages endpoint from a language without a mature provider SDK instead of the assigned second role because a standards-based HTTP client can send JSON requests with the correct headers and decode responses; their respective acceptance criteria are to reduce setup overhead when a service sends many short requests and to simplify authentication, serialization, and response typing in a Python service.

C:

The first proposal serves the different purpose to call the Messages endpoint from a language without a mature provider SDK instead of the assigned first role because a standards-based HTTP client can send JSON requests with the correct headers and decode responses, while the second proposal serves the different purpose to smooth sudden ingestion spikes before workers call a capacity-limited model API instead of the assigned second role because queueing decouples producers from consumer throughput and enables controlled retries; their respective acceptance criteria are to reduce setup overhead when a service sends many short requests and to simplify authentication, serialization, and response typing in a Python service.

D:

The first proposal serves the different purpose to smooth sudden ingestion spikes before workers call a capacity-limited model API instead of the assigned first role because queueing decouples producers from consumer throughput and enables controlled retries, while the second proposal fits its assigned role because maintained client libraries wrap the HTTP API and expose structured response objects; their respective acceptance criteria are to reduce setup overhead when a service sends many short requests and to simplify authentication, serialization, and response typing in a Python service.

Question 4

A team maintaining a backend integration needs to actually retrieve a customer's order status requested by Claude. Which implementation most directly satisfies the requirement?

  1. Read the HTTP error status before retrying
  2. Process tool_use responses in the application
  3. Call the Messages API with explicit model and max_tokens
  4. Supply a tool definition with input schema

Explanation

Correct Answer

B:

Client tools execute in application code; the model emits structured requests but does not invoke external functions itself; this behavior directly supports the need to actually retrieve a customer's order status requested by Claude. Displaying the tool request as the final answer never performs the lookup.

Incorrect Answers

A:

This approach serves the separate need to distinguish an invalid request from a temporary provider or rate-limit failure because status codes and structured errors guide correction versus bounded retry; it does not establish the outcome to actually retrieve a customer's order status requested by Claude, and blind retries of permanent validation errors waste time and quota.

C:

For the alternate objective to return an answer to a short user request through a backend service, a message request identifies the target model, a message list, and an output-token ceiling; the feature here must actually retrieve a customer's order status requested by Claude, and the API is not a stateful chat session unless prior turns are supplied.

D:

A declared tool name, description, and schema constrain the interface offered to the model; this applies to the different goal to let Claude request a read-only inventory lookup with typed arguments, not the requirement to actually retrieve a customer's order status requested by Claude; listing a function in a prompt does not create a callable tool contract.

Question 5

The acceptance criteria for an API reliability change state that it must prevent a bursty job service from exceeding allowed request throughput. Which implementation aligns with that criterion?

  1. Interpret max_tokens as a truncation cause
  2. Investigate a 401 authentication failure
  3. Investigate a 403 authorization failure
  4. Handle a 429 rate limit error

Explanation

Correct Answer

D:

A bounded retry policy and workload shaping address quota or rate throttling; this behavior directly supports the need to prevent a bursty job service from exceeding allowed request throughput. Repeating the request immediately can intensify the limit.

Incorrect Answers

A:

This approach serves the separate need to explain why a generated answer ends before a required final field because the stop reason indicates the generation budget was exhausted rather than the task being completed; it does not establish the outcome to prevent a bursty job service from exceeding allowed request throughput, and a valid parse of the partial prefix does not make the report complete.

B:

Authentication errors require checking credential validity and configuration; despite being useful when a service must restore an integration after an API key is revoked, it is not a solution to prevent a bursty job service from exceeding allowed request throughput; a larger output-token budget cannot authenticate the caller.

C:

For the alternate objective to determine why a valid service identity is denied a protected capability, authorization failures concern permission or policy after identity is established; the feature here must prevent a bursty job service from exceeding allowed request throughput, and rotating an otherwise valid key will not necessarily grant missing access.

Question 6

A team maintaining a retrieval pipeline needs to improve extraction quality after discovering consistent edge-case failures. Which implementation most directly satisfies the requirement?

  1. Use a stepwise decomposition
  2. Specify output format in the prompt
  3. Iterate prompts against an evaluation set
  4. Separate system policy from user content

Explanation

Correct Answer

C:

Controlled revisions measured on stable examples show whether a change actually helps; this behavior directly supports the need to improve extraction quality after discovering consistent edge-case failures. Optimizing against a single anecdote can degrade other cases.

Incorrect Answers

A:

This approach serves the separate need to solve a complex migration comparison that requires several distinct checks because breaking the work into verifiable stages lets the application check intermediate artifacts; it does not establish the outcome to improve extraction quality after discovering consistent edge-case failures, and simply increasing prompt length can make dependencies less clear.

B:

Explicit field descriptions guide generation toward the consumer's expected structure; despite being useful when a service must obtain named fields in a consistent response layout for subsequent processing, it is not a solution to improve extraction quality after discovering consistent edge-case failures; formatting guidance still requires machine validation at the boundary.

D:

Stable higher-priority instructions should be represented in the designated instruction channel; this applies to the different goal to prevent a customer message from overriding the application's escalation rules, not the requirement to improve extraction quality after discovering consistent edge-case failures; user text is the task data, not a privileged configuration surface.

Question 7

The acceptance criteria for an agent governance change state that it must prevent stack traces and debugging telemetry from disclosing tokens. Which implementation aligns with that criterion?

  1. Load keys from a managed secret service
  2. Use short-lived scoped credentials where supported
  3. Audit sensitive credential use
  4. Protect secrets in error traces

Explanation

Correct Answer

D:

Redaction at logging boundaries avoids accidentally copying credentials to broad observability systems; this behavior directly supports the need to prevent stack traces and debugging telemetry from disclosing tokens. Encrypted storage does not help when secrets are printed into logs.

Incorrect Answers

A:

Secrets managers support controlled retrieval, rotation, and access logs; this applies to the different goal to avoid committing provider credentials inside a shared repository, not the requirement to prevent stack traces and debugging telemetry from disclosing tokens; obscuring a literal string in source is not adequate key protection.

B:

When an application needs to limit the impact of a temporary worker credential being stolen, expiring identity grants reduce the time and permissions available to an attacker; the current capability must instead prevent stack traces and debugging telemetry from disclosing tokens, while permanent shared administrator keys expand blast radius.

C:

Usage and access logs support incident investigation and anomaly response; the supported outcome is to detect unexpected API key activity from an unfamiliar process, whereas this scenario requires the ability to prevent stack traces and debugging telemetry from disclosing tokens; a successful API call alone does not establish that use was authorized.

Question 8

Two separate tickets in a model platform backlog require different controls. One needs to design tests that tolerate benign wording variation while detecting semantic failures; the other needs to reduce variation when a service maps documents into consistent categories. Which implementation pair is correct?

  1. First component: Treat model outputs as stochastic. Second component: Use low-temperature generation for repeatable extraction.
  2. First component: Treat model outputs as stochastic. Second component: Recognize next-token generation.
  3. First component: Reduce reasoning effort on trivial routing. Second component: Use low-temperature generation for repeatable extraction.
  4. First component: Recognize next-token generation. Second component: Reduce reasoning effort on trivial routing.

Explanation

Correct Answer

A:

The first proposal fits its assigned role because non-determinism requires evaluation by criteria and repeated examples rather than exact-string equality everywhere, while the second proposal fits its assigned role because lower randomness often improves consistency for narrow, constrained tasks; their respective acceptance criteria are to design tests that tolerate benign wording variation while detecting semantic failures and to reduce variation when a service maps documents into consistent categories.

Incorrect Answers

B:

The first proposal fits its assigned role because non-determinism requires evaluation by criteria and repeated examples rather than exact-string equality everywhere, while the second proposal serves the different purpose to explain why a plausible answer can still contain an unsupported factual claim instead of the assigned second role because the model predicts likely continuations conditioned on context rather than querying an authoritative truth database by default; their respective acceptance criteria are to design tests that tolerate benign wording variation while detecting semantic failures and to reduce variation when a service maps documents into consistent categories.

C:

The first proposal serves the different purpose to shorten latency and spend on tasks that need only basic categorization instead of the assigned first role because lower effort can suit predictable work where deeper reasoning adds little measurable quality, while the second proposal fits its assigned role because lower randomness often improves consistency for narrow, constrained tasks; their respective acceptance criteria are to design tests that tolerate benign wording variation while detecting semantic failures and to reduce variation when a service maps documents into consistent categories.

D:

The first proposal serves the different purpose to explain why a plausible answer can still contain an unsupported factual claim instead of the assigned first role because the model predicts likely continuations conditioned on context rather than querying an authoritative truth database by default, while the second proposal serves the different purpose to shorten latency and spend on tasks that need only basic categorization instead of the assigned second role because lower effort can suit predictable work where deeper reasoning adds little measurable quality; their respective acceptance criteria are to design tests that tolerate benign wording variation while detecting semantic failures and to reduce variation when a service maps documents into consistent categories.

Question 9

The acceptance criteria for a production API change state that it must turn a vague request for an AI claims assistant into observable user actions and outputs. Which implementation aligns with that criterion?

  1. Document functional acceptance criteria
  2. Identify authoritative source systems
  3. Record nonfunctional latency targets
  4. Set an explicit data residency constraint
  5. Define failure and recovery objectives

Explanation

Correct Answer

A:

Acceptance criteria specify inputs, outputs, and successful behavior that can be tested before release; this behavior directly supports the need to turn a vague request for an AI claims assistant into observable user actions and outputs. Choosing a model first does not define what the application must accomplish.

Incorrect Answers

B:

When an application needs to prevent a Claude assistant from inventing account balances when it lacks current ledger data, source-of-truth mapping identifies which verified systems must supply facts and what freshness guarantees apply; the current capability must instead turn a vague request for an AI claims assistant into observable user actions and outputs, while model inference is not a substitute for a transactional database.

C:

A measurable response-time objective constrains interaction design and infrastructure choices; the supported outcome is to decide between streaming and offline processing for a time-sensitive support assistant, whereas this scenario requires the ability to turn a vague request for an AI claims assistant into observable user actions and outputs; functional prompts alone cannot establish acceptable performance.

D:

This approach serves the separate need to ensure protected records are processed only in approved jurisdictions because documented residency requirements inform cloud placement, logging, storage, and vendor selection; it does not establish the outcome to turn a vague request for an AI claims assistant into observable user actions and outputs, and a statement about answer quality does not control geographic data handling.

E:

Availability, recovery time, and fallback behavior must be specified as measurable operational needs; despite being useful when a service must keep a regulated triage service usable during an upstream API outage, it is not a solution to turn a vague request for an AI claims assistant into observable user actions and outputs; a happy-path prototype misses behavior under dependency failures.

Question 10

An engineer extending a data pipeline workflow has been asked to let Claude request a read-only inventory lookup with typed arguments. Which choice addresses the request most precisely?

  1. Handle message_stop and final usage events
  2. Supply a tool definition with input schema
  3. Process tool_use responses in the application
  4. Call the Messages API with explicit model and max_tokens
  5. Reconcile batch responses by custom_id

Explanation

Correct Answer

B:

A declared tool name, description, and schema constrain the interface offered to the model; this behavior directly supports the need to let Claude request a read-only inventory lookup with typed arguments. Listing a function in a prompt does not create a callable tool contract.

Incorrect Answers

A:

The final stream events signal completion and deliver accumulated metadata beyond individual text fragments; the supported outcome is to record complete streamed output and token usage accurately, whereas this scenario requires the ability to let Claude request a read-only inventory lookup with typed arguments; stopping after the first content delta truncates responses.

C:

Client tools execute in application code; the model emits structured requests but does not invoke external functions itself; despite being useful when a service must actually retrieve a customer's order status requested by Claude, it is not a solution to let Claude request a read-only inventory lookup with typed arguments; displaying the tool request as the final answer never performs the lookup.

D:

For the alternate objective to return an answer to a short user request through a backend service, a message request identifies the target model, a message list, and an output-token ceiling; the feature here must let Claude request a read-only inventory lookup with typed arguments, and the API is not a stateful chat session unless prior turns are supplied.

E:

A caller-assigned identifier links each result to the original request independently of result ordering; this applies to the different goal to match returned batch results to the right source documents despite varying completion order, not the requirement to let Claude request a read-only inventory lookup with typed arguments; assuming completion order matches submission order risks mixing customer data.

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