Anthropic CCDV-F (Claude Certified Developer - Foundations) Practice Test 49

Question 480

A release plan for a prompt engineering platform has two acceptance checks: component one must improve extraction quality after discovering consistent edge-case failures; component two must obtain named fields in a consistent response layout for subsequent processing. Which proposal passes both?

  1. First component: Iterate prompts against an evaluation set. Second component: State one measurable outcome per instruction.
  2. First component: State one measurable outcome per instruction. Second component: Separate system policy from user content.
  3. First component: Separate system policy from user content. Second component: Specify output format in the prompt.
  4. First component: Iterate prompts against an evaluation set. Second component: Specify output format in the prompt.

Explanation

Correct Answer

D:

The first proposal fits its assigned role because controlled revisions measured on stable examples show whether a change actually helps, while the second proposal fits its assigned role because explicit field descriptions guide generation toward the consumer's expected structure; their respective acceptance criteria are to improve extraction quality after discovering consistent edge-case failures and to obtain named fields in a consistent response layout for subsequent processing.

Incorrect Answers

A:

The first proposal fits its assigned role because controlled revisions measured on stable examples show whether a change actually helps, while the second proposal serves the different purpose to stop a summarizer from producing unrelated analysis instead of required action items instead of the assigned second role because clear task goals and acceptance criteria reduce competing interpretations; their respective acceptance criteria are to improve extraction quality after discovering consistent edge-case failures and to obtain named fields in a consistent response layout for subsequent processing.

B:

The first proposal serves the different purpose to stop a summarizer from producing unrelated analysis instead of required action items instead of the assigned first role because clear task goals and acceptance criteria reduce competing interpretations, while the second proposal serves the different purpose to prevent a customer message from overriding the application's escalation rules instead of the assigned second role because stable higher-priority instructions should be represented in the designated instruction channel; their respective acceptance criteria are to improve extraction quality after discovering consistent edge-case failures and to obtain named fields in a consistent response layout for subsequent processing.

C:

The first proposal serves the different purpose to prevent a customer message from overriding the application's escalation rules instead of the assigned first role because stable higher-priority instructions should be represented in the designated instruction channel, while the second proposal fits its assigned role because explicit field descriptions guide generation toward the consumer's expected structure; their respective acceptance criteria are to improve extraction quality after discovering consistent edge-case failures and to obtain named fields in a consistent response layout for subsequent processing.

Question 481

An MCP interoperability team owns two independent components. The first must prevent ambiguous generated arguments from reaching an inventory service; the second must prevent a helpful assistant from issuing refunds automatically. Which paired plan fulfills both?

  1. First component: Write task-specific tool descriptions. Second component: Restrict tool network credentials.
  2. First component: Define narrow tool input schemas. Second component: Write task-specific tool descriptions.
  3. First component: Restrict tool network credentials. Second component: Add an approval boundary before sensitive writes.
  4. First component: Validate generated arguments in application code. Second component: Write task-specific tool descriptions.
  5. First component: Define narrow tool input schemas. Second component: Add an approval boundary before sensitive writes.

Explanation

Correct Answer

E:

The first proposal fits its assigned role because typed parameter definitions constrain requests and enable strict validation before execution, while the second proposal fits its assigned role because authorization by a user or operator can be enforced before the executor commits changes; their respective acceptance criteria are to prevent ambiguous generated arguments from reaching an inventory service and to prevent a helpful assistant from issuing refunds automatically.

Incorrect Answers

A:

The first proposal serves the different purpose to let Claude distinguish a read-only invoice lookup from an invoice update instead of the assigned first role because clear capabilities and side effects guide tool selection without granting new authority, while the second proposal serves the different purpose to stop a directory lookup function from also gaining unrestricted database write access instead of the assigned second role because minimal capabilities reduce risk when generated arguments or retrieved text are hostile; their respective acceptance criteria are to prevent ambiguous generated arguments from reaching an inventory service and to prevent a helpful assistant from issuing refunds automatically.

B:

The first proposal fits its assigned role because typed parameter definitions constrain requests and enable strict validation before execution, while the second proposal serves the different purpose to let Claude distinguish a read-only invoice lookup from an invoice update instead of the assigned second role because clear capabilities and side effects guide tool selection without granting new authority; their respective acceptance criteria are to prevent ambiguous generated arguments from reaching an inventory service and to prevent a helpful assistant from issuing refunds automatically.

C:

The first proposal serves the different purpose to stop a directory lookup function from also gaining unrestricted database write access instead of the assigned first role because minimal capabilities reduce risk when generated arguments or retrieved text are hostile, while the second proposal fits its assigned role because authorization by a user or operator can be enforced before the executor commits changes; their respective acceptance criteria are to prevent ambiguous generated arguments from reaching an inventory service and to prevent a helpful assistant from issuing refunds automatically.

D:

The first proposal serves the different purpose to prevent malformed or unauthorized input from executing a customer data query instead of the assigned first role because the executor must check types, ranges, identity, and scope independently of the model, while the second proposal serves the different purpose to let Claude distinguish a read-only invoice lookup from an invoice update instead of the assigned second role because clear capabilities and side effects guide tool selection without granting new authority; their respective acceptance criteria are to prevent ambiguous generated arguments from reaching an inventory service and to prevent a helpful assistant from issuing refunds automatically.

Question 482

In a data pipeline project, acceptance test A checks whether a component can prevent a malicious help article from ordering the model to reveal secrets. Acceptance test B checks whether a different component can keep a downstream parser from crashing when a request returns no usable text block. Which plan satisfies both tests?

  1. First component: Use bounded retrieval results. Second component: Add cancellation handling to streaming UI.
  2. First component: Add cancellation handling to streaming UI. Second component: Handle empty and partial model responses.
  3. First component: Treat retrieval results as untrusted data. Second component: Use bounded retrieval results.
  4. First component: Treat retrieval results as untrusted data. Second component: Handle empty and partial model responses.

Explanation

Correct Answer

D:

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 content-aware response handling checks block types, stop reasons, and required values before proceeding; their respective acceptance criteria are to prevent a malicious help article from ordering the model to reveal secrets and to keep a downstream parser from crashing when a request returns no usable text block.

Incorrect Answers

A:

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 serves the different purpose to stop processing when a user cancels a long report and release resources cleanly instead of the assigned second role because cancellation signals should propagate through streams, queues, and workers where supported; their respective acceptance criteria are to prevent a malicious help article from ordering the model to reveal secrets and to keep a downstream parser from crashing when a request returns no usable text block.

B:

The first proposal serves the different purpose to stop processing when a user cancels a long report and release resources cleanly instead of the assigned first role because cancellation signals should propagate through streams, queues, and workers where supported, while the second proposal fits its assigned role because content-aware response handling checks block types, stop reasons, and required values before proceeding; their respective acceptance criteria are to prevent a malicious help article from ordering the model to reveal secrets and to keep a downstream parser from crashing when a request returns no usable text block.

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 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 keep a downstream parser from crashing when a request returns no usable text block.

Question 483

The acceptance criteria for a token economics change state that it must serve a mixed coding and support assistant at moderate latency and cost. Which implementation aligns with that criterion?

  1. Prefer a Sonnet-class model for balanced workloads
  2. Route hard cases to a stronger fallback
  3. Prefer a high-capability Opus-class model
  4. Check modality and feature support

Explanation

Correct Answer

A:

The balanced tier often offers a practical quality-throughput compromise, subject to task-level testing; this behavior directly supports the need to serve a mixed coding and support assistant at moderate latency and cost. It should not be assumed to dominate every model on every benchmark.

Incorrect Answers

B:

A tiered policy starts with a cheaper path and escalates on measurable uncertainty or validation failure; the supported outcome is to control average spend while retaining quality on exceptional requests, whereas this scenario requires the ability to serve a mixed coding and support assistant at moderate latency and cost; fixed use of the most costly model discards easy-case savings.

C:

This approach serves the separate need to handle a rare complex architecture review where reasoning accuracy outweighs cost and latency because the capability-oriented tier is suited to the hardest tasks after evaluation confirms an advantage; it does not establish the outcome to serve a mixed coding and support assistant at moderate latency and cost, and it is an inefficient default for every basic classification request.

D:

Availability of context length, structured output, and modalities may depend on the selected model and version; despite being useful when a service must avoid selecting a model that lacks a required document or tool feature, it is not a solution to serve a mixed coding and support assistant at moderate latency and cost; family-level generalizations do not override specific compatibility documentation.

Question 484

A production API team is evaluating two disjoint use cases. For one use case, it must avoid assuming a custom local database function runs inside Anthropic infrastructure; for the other, it must preserve a customer's follow-up reference to an earlier case number. Which paired choices are justified?

  1. First component: Differentiate client and server tool execution. Second component: Supply previous turns on a new API request.
  2. First component: Call the Messages API with explicit model and max_tokens. Second component: Supply previous turns on a new API request.
  3. First component: Handle message_stop and final usage events. Second component: Call the Messages API with explicit model and max_tokens.
  4. First component: Differentiate client and server tool execution. Second component: Handle message_stop and final usage events.
  5. First component: Process tool_use responses in the application. Second component: Handle message_stop and final usage events.

Explanation

Correct Answer

A:

The first proposal fits its assigned role because the integration owns execution of client tools, while supported server tools are performed by the provider, while the second proposal fits its assigned role because stateless requests need relevant history passed back in the messages sequence; their respective acceptance criteria are to avoid assuming a custom local database function runs inside Anthropic infrastructure and to preserve a customer's follow-up reference to an earlier case number.

Incorrect Answers

B:

The first proposal serves the different purpose to return an answer to a short user request through a backend service instead of the assigned first role because a message request identifies the target model, a message list, and an output-token ceiling, while the second proposal fits its assigned role because stateless requests need relevant history passed back in the messages sequence; their respective acceptance criteria are to avoid assuming a custom local database function runs inside Anthropic infrastructure and to preserve a customer's follow-up reference to an earlier case number.

C:

The first proposal serves the different purpose to record complete streamed output and token usage accurately instead of the assigned first role because the final stream events signal completion and deliver accumulated metadata beyond individual text fragments, while the second proposal serves the different purpose to return an answer to a short user request through a backend service instead of the assigned second role because a message request identifies the target model, a message list, and an output-token ceiling; their respective acceptance criteria are to avoid assuming a custom local database function runs inside Anthropic infrastructure and to preserve a customer's follow-up reference to an earlier case number.

D:

The first proposal fits its assigned role because the integration owns execution of client tools, while supported server tools are performed by the provider, while the second proposal serves the different purpose to record complete streamed output and token usage accurately instead of the assigned second role because the final stream events signal completion and deliver accumulated metadata beyond individual text fragments; their respective acceptance criteria are to avoid assuming a custom local database function runs inside Anthropic infrastructure and to preserve a customer's follow-up reference to an earlier case number.

E:

The first proposal serves the different purpose to actually retrieve a customer's order status requested by Claude instead of the assigned first role because client tools execute in application code; the model emits structured requests but does not invoke external functions itself, while the second proposal serves the different purpose to record complete streamed output and token usage accurately instead of the assigned second role because the final stream events signal completion and deliver accumulated metadata beyond individual text fragments; their respective acceptance criteria are to avoid assuming a custom local database function runs inside Anthropic infrastructure and to preserve a customer's follow-up reference to an earlier case number.

Question 485

A technical specification for a conversation design application assigns two jobs to separate modules: module A must prevent a customer-facing assistant from making final legal determinations without review, and module B must let the model summarize a hostile email without obeying its embedded commands. Which two-part solution is appropriate?

  1. First component: Specify the handling of missing information. Second component: Delimit untrusted source material.
  2. First component: Use explicit refusal and escalation instructions. Second component: Use a stepwise decomposition.
  3. First component: Use a stepwise decomposition. Second component: Specify the handling of missing information.
  4. First component: Use explicit refusal and escalation instructions. Second component: Delimit untrusted source material.

Explanation

Correct Answer

D:

The first proposal fits its assigned role because clear boundaries describe when to abstain or route to a qualified person, while the second proposal fits its assigned role because clear source boundaries and explicit treatment as quoted evidence help reduce instruction confusion; their respective acceptance criteria are to prevent a customer-facing assistant from making final legal determinations without review and to let the model summarize a hostile email without obeying its embedded commands.

Incorrect Answers

A:

The first proposal serves the different purpose to avoid invented contract dates when the uploaded document lacks them instead of the assigned first role because an explicit unknown or null rule makes uncertainty visible in the requested output, while the second proposal fits its assigned role because clear source boundaries and explicit treatment as quoted evidence help reduce instruction confusion; their respective acceptance criteria are to prevent a customer-facing assistant from making final legal determinations without review and to let the model summarize a hostile email without obeying its embedded commands.

B:

The first proposal fits its assigned role because clear boundaries describe when to abstain or route to a qualified person, while the second proposal serves the different purpose to solve a complex migration comparison that requires several distinct checks instead of the assigned second role because breaking the work into verifiable stages lets the application check intermediate artifacts; their respective acceptance criteria are to prevent a customer-facing assistant from making final legal determinations without review and to let the model summarize a hostile email without obeying its embedded commands.

C:

The first proposal serves the different purpose to solve a complex migration comparison that requires several distinct checks instead of the assigned first role because breaking the work into verifiable stages lets the application check intermediate artifacts, while the second proposal serves the different purpose to avoid invented contract dates when the uploaded document lacks them instead of the assigned second role because an explicit unknown or null rule makes uncertainty visible in the requested output; their respective acceptance criteria are to prevent a customer-facing assistant from making final legal determinations without review and to let the model summarize a hostile email without obeying its embedded commands.

Question 486

After pilot feedback, a data pipeline group needs two independent improvements. The first is to maintain multiple Claude entry points without drifting retry and logging behavior; the second is to debug where a request stalls between the web UI, queue, and Claude gateway. Which engineering proposal handles both?

  1. First component: Refactor duplicated request orchestration. Second component: Enforce concurrency limits.
  2. First component: Enforce concurrency limits. Second component: Use exponential backoff with jitter.
  3. First component: Use exponential backoff with jitter. Second component: Correlate distributed traces across services.
  4. First component: Refactor duplicated request orchestration. Second component: Correlate distributed traces across services.

Explanation

Correct Answer

D:

The first proposal fits its assigned role because a shared, tested client layer centralizes cross-cutting concerns and reduces inconsistent fixes, while the second proposal fits its assigned role because a trace identifier links timing and errors through multiple process boundaries; their respective acceptance criteria are to maintain multiple Claude entry points without drifting retry and logging behavior and to debug where a request stalls between the web UI, queue, and Claude gateway.

Incorrect Answers

A:

The first proposal fits its assigned role because a shared, tested client layer centralizes cross-cutting concerns and reduces inconsistent fixes, while the second proposal serves the different purpose to prevent bursts of document jobs from overwhelming rate limits and memory instead of the assigned second role because a semaphore or bounded queue keeps in-flight work within measured capacity; their respective acceptance criteria are to maintain multiple Claude entry points without drifting retry and logging behavior and to debug where a request stalls between the web UI, queue, and Claude gateway.

B:

The first proposal serves the different purpose to prevent bursts of document jobs from overwhelming rate limits and memory instead of the assigned first role because a semaphore or bounded queue keeps in-flight work within measured capacity, while the second proposal serves the different purpose to recover from transient 429 responses without synchronized retry storms instead of the assigned second role because increasing randomized delays reduce contention and respect service throttling signals; their respective acceptance criteria are to maintain multiple Claude entry points without drifting retry and logging behavior and to debug where a request stalls between the web UI, queue, and Claude gateway.

C:

The first proposal serves the different purpose to recover from transient 429 responses without synchronized retry storms instead of the assigned first role because increasing randomized delays reduce contention and respect service throttling signals, while the second proposal fits its assigned role because a trace identifier links timing and errors through multiple process boundaries; their respective acceptance criteria are to maintain multiple Claude entry points without drifting retry and logging behavior and to debug where a request stalls between the web UI, queue, and Claude gateway.

Question 487

A technical specification for an engineering productivity application assigns two jobs to separate modules: module A must coordinate separate billing, compliance, and engineering analyses under one final decision, and module B must execute an unchanged sequence of extraction, validation, and publishing steps. Which two-part solution is appropriate?

  1. First component: Pass structured handoff artifacts between workers. Second component: Use an agentic decision loop.
  2. First component: Adopt a supervisor with specialist workers. Second component: Use a deterministic workflow.
  3. First component: Use an agentic decision loop. Second component: Use a deterministic workflow.
  4. First component: Adopt a supervisor with specialist workers. Second component: Pass structured handoff artifacts between workers.

Explanation

Correct Answer

B:

The first proposal fits its assigned role because a managing agent delegates bounded subtasks to specialists and reconciles their outputs against a common objective, while the second proposal fits its assigned role because a fixed pipeline keeps control flow in application code, making execution order reproducible and easy to audit; their respective acceptance criteria are to coordinate separate billing, compliance, and engineering analyses under one final decision and to execute an unchanged sequence of extraction, validation, and publishing steps.

Incorrect Answers

A:

The first proposal serves the different purpose to preserve evidence and data lineage when one agent delegates a subtask to another instead of the assigned first role because defined input and output schemas make it possible to attribute conclusions and detect missing fields at the handoff, while the second proposal serves the different purpose to investigate an open-ended incident where the next query depends on earlier findings instead of the assigned second role because a model-directed loop can choose subsequent actions based on tool observations until it reaches a stopping condition; their respective acceptance criteria are to coordinate separate billing, compliance, and engineering analyses under one final decision and to execute an unchanged sequence of extraction, validation, and publishing steps.

C:

The first proposal serves the different purpose to investigate an open-ended incident where the next query depends on earlier findings instead of the assigned first role because a model-directed loop can choose subsequent actions based on tool observations until it reaches a stopping condition, while the second proposal fits its assigned role because a fixed pipeline keeps control flow in application code, making execution order reproducible and easy to audit; their respective acceptance criteria are to coordinate separate billing, compliance, and engineering analyses under one final decision and to execute an unchanged sequence of extraction, validation, and publishing steps.

D:

The first proposal fits its assigned role because a managing agent delegates bounded subtasks to specialists and reconciles their outputs against a common objective, while the second proposal serves the different purpose to preserve evidence and data lineage when one agent delegates a subtask to another instead of the assigned second role because defined input and output schemas make it possible to attribute conclusions and detect missing fields at the handoff; their respective acceptance criteria are to coordinate separate billing, compliance, and engineering analyses under one final decision and to execute an unchanged sequence of extraction, validation, and publishing steps.

Question 488

A conversation design team owns two independent components. The first must prevent a customer-facing assistant from making final legal determinations without review; the second must prevent a customer message from overriding the application's escalation rules. Which paired plan fulfills both?

  1. First component: Use explicit refusal and escalation instructions. Second component: Specify the handling of missing information.
  2. First component: Use explicit refusal and escalation instructions. Second component: Separate system policy from user content.
  3. First component: Specify the handling of missing information. Second component: Use a stepwise decomposition.
  4. First component: Use a stepwise decomposition. Second component: Separate system policy from user content.
  5. First component: State one measurable outcome per instruction. Second component: Specify the handling of missing information.

Explanation

Correct Answer

B:

The first proposal fits its assigned role because clear boundaries describe when to abstain or route to a qualified person, while the second proposal fits its assigned role because stable higher-priority instructions should be represented in the designated instruction channel; their respective acceptance criteria are to prevent a customer-facing assistant from making final legal determinations without review and to prevent a customer message from overriding the application's escalation rules.

Incorrect Answers

A:

The first proposal fits its assigned role because clear boundaries describe when to abstain or route to a qualified person, while the second proposal serves the different purpose to avoid invented contract dates when the uploaded document lacks them instead of the assigned second role because an explicit unknown or null rule makes uncertainty visible in the requested output; their respective acceptance criteria are to prevent a customer-facing assistant from making final legal determinations without review and to prevent a customer message from overriding the application's escalation rules.

C:

The first proposal serves the different purpose to avoid invented contract dates when the uploaded document lacks them instead of the assigned first role because an explicit unknown or null rule makes uncertainty visible in the requested output, while the second proposal serves the different purpose to solve a complex migration comparison that requires several distinct checks instead of the assigned second role because breaking the work into verifiable stages lets the application check intermediate artifacts; their respective acceptance criteria are to prevent a customer-facing assistant from making final legal determinations without review and to prevent a customer message from overriding the application's escalation rules.

D:

The first proposal serves the different purpose to solve a complex migration comparison that requires several distinct checks instead of the assigned first role because breaking the work into verifiable stages lets the application check intermediate artifacts, while the second proposal fits its assigned role because stable higher-priority instructions should be represented in the designated instruction channel; their respective acceptance criteria are to prevent a customer-facing assistant from making final legal determinations without review and to prevent a customer message from overriding the application's escalation rules.

E:

The first proposal serves the different purpose to stop a summarizer from producing unrelated analysis instead of required action items instead of the assigned first role because clear task goals and acceptance criteria reduce competing interpretations, while the second proposal serves the different purpose to avoid invented contract dates when the uploaded document lacks them instead of the assigned second role because an explicit unknown or null rule makes uncertainty visible in the requested output; their respective acceptance criteria are to prevent a customer-facing assistant from making final legal determinations without review and to prevent a customer message from overriding the application's escalation rules.

Question 489

An engineer extending a production API workflow has been asked to recover a response that ended because its output budget was exhausted. Which choice addresses the request most precisely?

  1. Use supported document or PDF input
  2. Handle the max_tokens stop_reason
  3. Send tool_result content back to Claude
  4. Reconcile batch responses by custom_id
  5. Process tool_use responses in the application

Explanation

Correct Answer

B:

The finish reason identifies truncation, so a caller may need a continuation or a revised token budget; this behavior directly supports the need to recover a response that ended because its output budget was exhausted. Treating that response as a complete answer can hide missing fields.

Incorrect Answers

A:

Document-aware input lets a capable model consider page content and graphics when supported; this applies to the different goal to extract claims from a document while retaining its visual layout, not the requirement to recover a response that ended because its output budget was exhausted; pasting only extracted text can discard useful visual context.

C:

A result block tied to the tool call returns the external observation to the model for synthesis; the supported outcome is to produce a user-facing answer after a successful inventory lookup, whereas this scenario requires the ability to recover a response that ended because its output budget was exhausted; a function's return value in server memory is invisible until included in the next request.

D:

This approach serves the separate need to match returned batch results to the right source documents despite varying completion order because a caller-assigned identifier links each result to the original request independently of result ordering; it does not establish the outcome to recover a response that ended because its output budget was exhausted, and assuming completion order matches submission order risks mixing customer data.

E:

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 recover a response that ended because its output budget was exhausted; displaying the tool request as the final answer never performs the lookup.

Question 490

The current design of a token economics pipeline does not adequately lower spend for large scheduled classification jobs that do not need immediate responses. Which improvement resolves the gap?

  1. Set maximum step counts for agents
  2. Track input and output usage separately
  3. Use batch pricing where deadline permits
  4. Measure cache read tokens

Explanation

Correct Answer

C:

Asynchronous batch processing can offer a cost-effective throughput path under supported terms; this behavior directly supports the need to lower spend for large scheduled classification jobs that do not need immediate responses. A night job should not be optimized with live streaming unless its users need it.

Incorrect Answers

A:

For the alternate objective to contain cumulative token and tool cost across a recursive research session, a bound on iterations prevents repeated generation and calls from growing without limit; the feature here must lower spend for large scheduled classification jobs that do not need immediate responses, and a per-message output cap alone does not control total multi-step spend.

B:

Metering by token category reveals the main cost driver and supports realistic forecasts; this applies to the different goal to forecast cost for an application that sends long documents but returns brief classifications, not the requirement to lower spend for large scheduled classification jobs that do not need immediate responses; counting only request totals hides prompt-size variability.

D:

Usage fields show how much eligible cached input was read rather than recomputed; the supported outcome is to verify that an optimization actually reuses an earlier prompt prefix, whereas this scenario requires the ability to lower spend for large scheduled classification jobs that do not need immediate responses; assuming identical UI prompts means a cache hit can conceal changed metadata or order.

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