Anthropic CCA-F (Claude Certified Architect - Foundations (CCAR-F)) Practice Test 53

Question 520

The release checklist for a tender-response data extractor explicitly checks that a partially failed batch is repaired efficiently; it also checks that an uncertain high-impact decision is routed for review instead of automatic execution. What coordinated solution meets the checklist?

  1. rerun the entire completed batch whenever one record fails; and proceed whenever the assistant's prose sounds certain
  2. rerun the entire completed batch whenever one record fails; and set a clear evidence or confidence boundary and require review below it
  3. identify failed records and resubmit only safe retry candidates; and set a clear evidence or confidence boundary and require review below it
  4. identify failed records and resubmit only safe retry candidates; and proceed whenever the assistant's prose sounds certain

Explanation

Correct Answer

C:

In a tender-response data extractor, selective retry avoids paying again for already successful items; alongside this, risk-weighted gates limit autonomous actions when support is inadequate, satisfying the distinct requirements that a partially failed batch is repaired efficiently and that an uncertain high-impact decision is routed for review instead of automatic execution.

Incorrect Answers

A:

In a tender-response data extractor, successful work would be duplicated unnecessarily and fluent language is not a calibrated reliability measure; the design would fail both conditions: a partially failed batch is repaired efficiently and an uncertain high-impact decision is routed for review instead of automatic execution.

B:

In a tender-response data extractor, risk-weighted gates limit autonomous actions when support is inadequate, but successful work would be duplicated unnecessarily; therefore the separate requirement that a partially failed batch is repaired efficiently would remain unsatisfied.

D:

In a tender-response data extractor, selective retry avoids paying again for already successful items, but fluent language is not a calibrated reliability measure; therefore the separate requirement that an uncertain high-impact decision is routed for review instead of automatic execution would remain unsatisfied.

Question 521

A team operating a CLI application maintenance task wants to establish two separate behaviors: a team shares a repository-specific MCP server configuration; in addition, tool output is normalized before it enters the agent context. Which plan implements both?

  1. commit a reviewed project .mcp.json without credentials; and use a PostToolUse processing hook to standardize the returned data
  2. commit a reviewed project .mcp.json without credentials; and preprocess only the original user message
  3. configure each contributor's personal MCP registry manually; and preprocess only the original user message
  4. configure each contributor's personal MCP registry manually; and use a PostToolUse processing hook to standardize the returned data

Explanation

Correct Answer

A:

In a CLI application maintenance task, project-scoped configuration can be distributed with the repository while keeping secrets external; alongside this, post-execution interception can clean result fields and attach useful status without changing the model's tool selection, satisfying the distinct requirements that a team shares a repository-specific MCP server configuration and that tool output is normalized before it enters the agent context.

Incorrect Answers

B:

In a CLI application maintenance task, project-scoped configuration can be distributed with the repository while keeping secrets external, but input normalization does not correct inconsistent tool outputs; therefore the separate requirement that tool output is normalized before it enters the agent context would remain unsatisfied.

C:

In a CLI application maintenance task, local personal settings do not create a reproducible team project setup and input normalization does not correct inconsistent tool outputs; the design would fail both conditions: a team shares a repository-specific MCP server configuration and tool output is normalized before it enters the agent context.

D:

In a CLI application maintenance task, post-execution interception can clean result fields and attach useful status without changing the model's tool selection, but local personal settings do not create a reproducible team project setup; therefore the separate requirement that a team shares a repository-specific MCP server configuration would remain unsatisfied.

Question 522

Tests for a security-advisory aggregation workflow will pass only if the model can distinguish two similar business tools and if workload growth does not cause uncontrolled parallel fan-out. Which proposed design directly satisfies both checks?

  1. give each tool a specific purpose and exclusion examples in its description; and launch all possible child tasks at once without resource controls
  2. give both tools vague names with identical generic descriptions; and limit concurrency and schedule independent jobs against execution budgets
  3. give both tools vague names with identical generic descriptions; and launch all possible child tasks at once without resource controls
  4. give each tool a specific purpose and exclusion examples in its description; and limit concurrency and schedule independent jobs against execution budgets

Explanation

Correct Answer

D:

In a security-advisory aggregation workflow, discriminative descriptions reduce wrong routing when operations sound alike; alongside this, bounded orchestration balances throughput with tool quotas and downstream service capacity, satisfying the distinct requirements that the model can distinguish two similar business tools and that workload growth does not cause uncontrolled parallel fan-out.

Incorrect Answers

A:

In a security-advisory aggregation workflow, discriminative descriptions reduce wrong routing when operations sound alike, but unbounded fan-out can overwhelm tools and blow through cost budgets; therefore the separate requirement that workload growth does not cause uncontrolled parallel fan-out would remain unsatisfied.

B:

In a security-advisory aggregation workflow, bounded orchestration balances throughput with tool quotas and downstream service capacity, but a model cannot reliably select on boundaries that the interface never explains; therefore the separate requirement that the model can distinguish two similar business tools would remain unsatisfied.

C:

In a security-advisory aggregation workflow, a model cannot reliably select on boundaries that the interface never explains and unbounded fan-out can overwhelm tools and blow through cost budgets; the design would fail both conditions: the model can distinguish two similar business tools and workload growth does not cause uncontrolled parallel fan-out.

Question 523

An architecture review of a legacy service decomposition finds two missing guarantees: a code investigation finds where a function is referenced; a team compares two candidate patches against the same baseline. What two-part change closes both gaps?

  1. ask for broad prose summaries of every package first; and choose whichever patch has the more persuasive natural-language summary
  2. use Grep to locate occurrences before opening focused files; and evaluate alternatives using the same tests and scoped diffs
  3. use Grep to locate occurrences before opening focused files; and choose whichever patch has the more persuasive natural-language summary
  4. ask for broad prose summaries of every package first; and evaluate alternatives using the same tests and scoped diffs

Explanation

Correct Answer

B:

In a legacy service decomposition, symbol searching narrows the evidence to files that mention the target; alongside this, consistent acceptance evidence reveals which candidate improves the relevant behavior, satisfying the distinct requirements that a code investigation finds where a function is referenced and that a team compares two candidate patches against the same baseline.

Incorrect Answers

A:

In a legacy service decomposition, high-level summaries can miss the exact call site and rhetorical fluency does not establish code correctness; the design would fail both conditions: a code investigation finds where a function is referenced and a team compares two candidate patches against the same baseline.

C:

In a legacy service decomposition, symbol searching narrows the evidence to files that mention the target, but rhetorical fluency does not establish code correctness; therefore the separate requirement that a team compares two candidate patches against the same baseline would remain unsatisfied.

D:

In a legacy service decomposition, consistent acceptance evidence reveals which candidate improves the relevant behavior, but high-level summaries can miss the exact call site; therefore the separate requirement that a code investigation finds where a function is referenced would remain unsatisfied.

Question 524

While preparing a procurement research assistant for deployment, engineers must ensure that a complex document receives focused quality checks; they must also ensure that a tool receives a typed and bounded request. Which combination best satisfies those requirements?

  1. use one generic request to make the document better; and define required properties and types in the input schema and validate arguments at the execution boundary
  2. separate factual, structural, and policy review passes; and define required properties and types in the input schema and validate arguments at the execution boundary
  3. use one generic request to make the document better; and accept one unstructured free-text argument for every operation
  4. separate factual, structural, and policy review passes; and accept one unstructured free-text argument for every operation

Explanation

Correct Answer

B:

In a procurement research assistant, specialized criteria can detect different failure modes without one overloaded evaluation instruction; alongside this, input schemas describe legal arguments, while explicit validation prevents unsupported values from reaching tool logic, satisfying the distinct requirements that a complex document receives focused quality checks and that a tool receives a typed and bounded request.

Incorrect Answers

A:

In a procurement research assistant, input schemas describe legal arguments, while explicit validation prevents unsupported values from reaching tool logic, but an undifferentiated review can overlook specific required constraints; therefore the separate requirement that a complex document receives focused quality checks would remain unsatisfied.

C:

In a procurement research assistant, an undifferentiated review can overlook specific required constraints and unconstrained prose leaves parsing and validation errors to ad hoc handler inference; the design would fail both conditions: a complex document receives focused quality checks and a tool receives a typed and bounded request.

D:

In a procurement research assistant, specialized criteria can detect different failure modes without one overloaded evaluation instruction, but unconstrained prose leaves parsing and validation errors to ad hoc handler inference; therefore the separate requirement that a tool receives a typed and bounded request would remain unsatisfied.

Question 525

Production readiness for a developer codebase exploration assistant depends on ensuring that client-side tool requests trigger another model turn after execution and that a missing MCP capability is diagnosed before changing prompts. Select the most appropriate combined approach.

  1. branch on stop_reason tool_use and execute returned tool_use blocks; and verify server connection, transport, and exposed tool availability
  2. end the loop whenever assistant text appears; and verify server connection, transport, and exposed tool availability
  3. end the loop whenever assistant text appears; and increase the system prompt length to compensate for disconnected tools
  4. branch on stop_reason tool_use and execute returned tool_use blocks; and increase the system prompt length to compensate for disconnected tools

Explanation

Correct Answer

A:

In a developer codebase exploration assistant, the response explicitly signals pending client execution and the loop must provide results before another model decision; alongside this, tool discovery depends on a live configured integration, not just the prompt wording, satisfying the distinct requirements that client-side tool requests trigger another model turn after execution and that a missing MCP capability is diagnosed before changing prompts.

Incorrect Answers

B:

In a developer codebase exploration assistant, tool discovery depends on a live configured integration, not just the prompt wording, but text may accompany a tool request and is not the completion signal; therefore the separate requirement that client-side tool requests trigger another model turn after execution would remain unsatisfied.

C:

In a developer codebase exploration assistant, text may accompany a tool request and is not the completion signal and a model cannot call a capability absent from its available tool registry; the design would fail both conditions: client-side tool requests trigger another model turn after execution and a missing MCP capability is diagnosed before changing prompts.

D:

In a developer codebase exploration assistant, the response explicitly signals pending client execution and the loop must provide results before another model decision, but a model cannot call a capability absent from its available tool registry; therefore the separate requirement that a missing MCP capability is diagnosed before changing prompts would remain unsatisfied.

Question 526

While preparing a supplier invoice extractor for deployment, engineers must ensure that a consequential claim receives independent human sign-off; they must also ensure that persistent ambiguous data receives human review. Which combination best satisfies those requirements?

  1. route high-impact or low-evidence results to an authorized reviewer; and keep retrying until the model finally emits something that passes
  2. accept model confidence as a substitute for authorization; and keep retrying until the model finally emits something that passes
  3. route high-impact or low-evidence results to an authorized reviewer; and stop after defined retries and route unresolved cases to a reviewer
  4. accept model confidence as a substitute for authorization; and stop after defined retries and route unresolved cases to a reviewer

Explanation

Correct Answer

C:

In a supplier invoice extractor, human approval supplies accountability where a mistaken model result has material cost; alongside this, a bounded escalation path prevents infinite loops and invented resolutions, satisfying the distinct requirements that a consequential claim receives independent human sign-off and that persistent ambiguous data receives human review.

Incorrect Answers

A:

In a supplier invoice extractor, human approval supplies accountability where a mistaken model result has material cost, but unlimited retries waste resources and can reinforce guesses; therefore the separate requirement that persistent ambiguous data receives human review would remain unsatisfied.

B:

In a supplier invoice extractor, a model score does not confer business or legal authority and unlimited retries waste resources and can reinforce guesses; the design would fail both conditions: a consequential claim receives independent human sign-off and persistent ambiguous data receives human review.

D:

In a supplier invoice extractor, a bounded escalation path prevents infinite loops and invented resolutions, but a model score does not confer business or legal authority; therefore the separate requirement that a consequential claim receives independent human sign-off would remain unsatisfied.

Question 527

During review of a procurement document processing system, an unmet criterion states that temporary service faults can be retried selectively. Which measure resolves this specific gap?

  1. return an explicit partial status with missing-item identifiers
  2. check operation status or use an idempotency key before retry
  3. retry every error with the same request indefinitely
  4. include a retryable classification and bounded retry advice

Explanation

Correct Answer

D:

For a procurement document processing system, typed transient failures allow backoff without treating permanent errors the same way; this directly establishes that temporary service faults can be retried selectively instead of relying on an unrelated workflow control.

Incorrect Answers

A:

Although failure metadata preserves the difference between a sparse valid response and incomplete execution, that only supports the distinct goal that partial data is not presented as a complete successful result, not the requirement that temporary service faults can be retried selectively in a procurement document processing system.

B:

Although a timeout can occur after the server commits the action, so retry safety needs state checks, that only supports the distinct goal that ambiguous write timeouts do not trigger duplicate side effects, not the requirement that temporary service faults can be retried selectively in a procurement document processing system.

C:

In a procurement document processing system, permission and validation failures will not heal through repetition; as a result, the stated requirement that temporary service faults can be retried selectively would still not be met.

Question 528

Before expanding a repository release validation process, the owner must demonstrate that a broad tool catalog does not overwhelm a small task, while also demonstrating that a straightforward lookup avoids unnecessary agent coordination. Which design provides both controls?

  1. load all available MCP tools into every specialist turn; and spawn a coordinator plus multiple specialists for every trivial query
  2. load all available MCP tools into every specialist turn; and use one focused tool-assisted agent for the bounded task
  3. supply only context-relevant tools for the current agent; and use one focused tool-assisted agent for the bounded task
  4. supply only context-relevant tools for the current agent; and spawn a coordinator plus multiple specialists for every trivial query

Explanation

Correct Answer

C:

In a repository release validation process, reducing irrelevant tool definitions improves selection and saves context; alongside this, orchestration overhead is not justified when there is no useful specialization or concurrency, satisfying the distinct requirements that a broad tool catalog does not overwhelm a small task and that a straightforward lookup avoids unnecessary agent coordination.

Incorrect Answers

A:

In a repository release validation process, tool bloat makes the available actions harder to discriminate and extra agents add latency and handoff risk without new capabilities; the design would fail both conditions: a broad tool catalog does not overwhelm a small task and a straightforward lookup avoids unnecessary agent coordination.

B:

In a repository release validation process, orchestration overhead is not justified when there is no useful specialization or concurrency, but tool bloat makes the available actions harder to discriminate; therefore the separate requirement that a broad tool catalog does not overwhelm a small task would remain unsatisfied.

D:

In a repository release validation process, reducing irrelevant tool definitions improves selection and saves context, but extra agents add latency and handoff risk without new capabilities; therefore the separate requirement that a straightforward lookup avoids unnecessary agent coordination would remain unsatisfied.

Question 529

An incident postmortem for a contract clause extraction workflow requires controls so that a selector accepts only supported operation names and that tool responses are associated with the exact invocations that produced them. Which paired remedy covers both failure modes?

  1. ask the model to invent a suitable action string; and return tool results with newly invented identifiers
  2. declare an operation enum in the schema and reject values outside it in the tool handler; and return tool_result blocks using the original tool_use_id values
  3. declare an operation enum in the schema and reject values outside it in the tool handler; and return tool results with newly invented identifiers
  4. ask the model to invent a suitable action string; and return tool_result blocks using the original tool_use_id values

Explanation

Correct Answer

B:

In a contract clause extraction workflow, schema constraints guide valid selections and runtime validation blocks unsupported operation values; alongside this, correlation identifiers bind asynchronous or parallel results to the right requests, satisfying the distinct requirements that a selector accepts only supported operation names and that tool responses are associated with the exact invocations that produced them.

Incorrect Answers

A:

In a contract clause extraction workflow, free-form labels can name unimplemented operations, and a prompt is not an enforced allowlist and the model cannot reliably associate a result with an outstanding call; the design would fail both conditions: a selector accepts only supported operation names and tool responses are associated with the exact invocations that produced them.

C:

In a contract clause extraction workflow, schema constraints guide valid selections and runtime validation blocks unsupported operation values, but the model cannot reliably associate a result with an outstanding call; therefore the separate requirement that tool responses are associated with the exact invocations that produced them would remain unsatisfied.

D:

In a contract clause extraction workflow, correlation identifiers bind asynchronous or parallel results to the right requests, but free-form labels can name unimplemented operations, and a prompt is not an enforced allowlist; therefore the separate requirement that a selector accepts only supported operation names would remain unsatisfied.

Question 530

The release checklist for a service-credit approval assistant explicitly checks that persistent ambiguous data receives human review; it also checks that a classifier avoids marking uncertain cases as confirmed violations. What coordinated solution meets the checklist?

  1. keep retrying until the model finally emits something that passes; and tell the model to flag anything that sounds suspicious
  2. keep retrying until the model finally emits something that passes; and define explicit positive evidence and an abstain or review category
  3. stop after defined retries and route unresolved cases to a reviewer; and define explicit positive evidence and an abstain or review category
  4. implement only stop after defined retries and route unresolved cases to a reviewer; make no corresponding change for the second requirement
  5. stop after defined retries and route unresolved cases to a reviewer; and tell the model to flag anything that sounds suspicious

Explanation

Correct Answer

C:

In a service-credit approval assistant, a bounded escalation path prevents infinite loops and invented resolutions; alongside this, clear decision thresholds separate direct evidence from weak similarity, satisfying the distinct requirements that persistent ambiguous data receives human review and that a classifier avoids marking uncertain cases as confirmed violations.

Incorrect Answers

A:

In a service-credit approval assistant, unlimited retries waste resources and can reinforce guesses and vague semantic similarity raises false positives; the design would fail both conditions: persistent ambiguous data receives human review and a classifier avoids marking uncertain cases as confirmed violations.

B:

In a service-credit approval assistant, clear decision thresholds separate direct evidence from weak similarity, but unlimited retries waste resources and can reinforce guesses; therefore the separate requirement that persistent ambiguous data receives human review would remain unsatisfied.

D:

In a service-credit approval assistant, a bounded escalation path prevents infinite loops and invented resolutions, but the independent mechanism for the other criterion is absent: clear decision thresholds separate direct evidence from weak similarity; as written, the design does not ensure that a classifier avoids marking uncertain cases as confirmed violations.

E:

In a service-credit approval assistant, a bounded escalation path prevents infinite loops and invented resolutions, but vague semantic similarity raises false positives; therefore the separate requirement that a classifier avoids marking uncertain cases as confirmed violations would remain unsatisfied.

Claude Certified Architect - Foundations (CCAR-F) (CCA-F) Practice Tests