Anthropic CCA-F (Claude Certified Architect - Foundations (CCAR-F)) Practice Test 2

Question 10

A long automation run applies several patches without making checkpoints. What practice improves recovery?

  1. Review and checkpoint cohesive changes with their test evidence
  2. Inspect the actual dependency version and local interface before rewriting calls
  3. Introduce the failing edge case to tests and implement the smallest correction
  4. Collect reproduction inputs and compare observed environment differences first

Explanation

Correct Answer

A:

Small validated increments provide rollback boundaries and clearer blame history. A giant unverified diff makes failures difficult to isolate.

Incorrect Answers

B:

Version drift can make a plausible call invalid; that serves aligning generated code to installed dependency APIs, whereas the case requires checkpointing incremental implementation progress.

C:

Because treating review feedback as a concrete test turns a discovered omission into durable coverage, the measure supports incorporating edge-case feedback into tests rather than the needed checkpointing incremental implementation progress.

D:

Without a reproducible failure or causal evidence, broad edits are speculative; its benefits for establishing a reliable bug reproduction do not resolve the need for checkpointing incremental implementation progress.

Question 11

A pipeline reviews an entire monorepo on every one-line change and consumes excessive tokens. What optimization fits?

  1. Ask for machine-readable JSON output and validate its schema
  2. Mandate file path, line range and finding rationale in a validated result schema
  3. Execute Claude Code non-interactively with claude -p in CI
  4. Scope retrieval and review to the changed files and their affected dependencies
  5. Specify explicit severity criteria and require evidence for blocking findings
  6. Provide relevant migration files and compatibility criteria with the diff

Explanation

Correct Answer

D:

A diff-aware prompt and bounded context reduce cost while preserving relevant evidence. Blindly scanning unrelated code dilutes signal and increases latency.

Incorrect Answers

A:

Structured output is easier for CI to parse and gate deterministically; this may support parsing review output safely in CI but cannot alone meet the requirement for constraining CI review to changed scope.

B:

Location-aware structured findings allow a bot to attach actionable comments; the stated problem requires constraining CI review to changed scope rather than generating location-specific PR findings.

C:

The print-style command supports scripting without a chat session; a method suited to invoking Claude Code in a noninteractive pipeline does not meet the need for constraining CI review to changed scope.

E:

Reviewer outputs should distinguish verified blockers from optional recommendations; that may serve calibrating automated code-review severity, but it is no substitute for constraining CI review to changed scope.

F:

Reviewing only altered lines can miss deployment sequencing and consumers; this may support providing dependency context for migration review but cannot alone meet the requirement for constraining CI review to changed scope.

Question 12

Multiple support agents must not advance the same session simultaneously with conflicting operations. What orchestration rule helps?

  1. Recheck identity and permissions before privileged continuation
  2. Verify the persisted operation ID against the payment system before retrying
  3. Serialize state-changing updates or use optimistic concurrency on the session
  4. Begin a separate session with its own customer and authorization context

Explanation

Correct Answer

C:

Concurrent writes can produce lost updates and inconsistent tool state. A version check or lock lets one ordered session history remain authoritative.

Incorrect Answers

A:

A stored conversation is not a perpetual permission grant; that relates to revalidating authorization on session resume, not the necessary preventing concurrent session-state conflicts.

B:

Although a durable transcript does not guarantee whether the external effect committed, the choice addresses reconciling external actions after session interruption and not the requested preventing concurrent session-state conflicts.

D:

Separate sessions prevent accidental disclosure and policy contamination between customers; the option would serve isolating unrelated customer conversations, not preventing concurrent session-state conflicts as specified in the stem.

Question 13

A security policy forbids shell execution, yet the task only requires reading a JSON file. What built-in choice avoids unnecessary privilege?

  1. Choose Grep to locate matching text across the repository
  2. Use Read rather than invoking Bash to print the file
  3. Find with Glob the matching migration paths and then inspect them in order
  4. Use Grep to search the symbol and then Read the matching files
  5. Choose Glob to enumerate files matching the path pattern

Explanation

Correct Answer

B:

Reading file content through the dedicated tool avoids an unrestricted shell surface. Matching the tool to the operation supports least privilege and clearer audit traces.

Incorrect Answers

A:

Grep searches contents and identifies files or lines containing a pattern; in this scenario, finding a phrase inside repository files is secondary to choosing direct file inspection over shell execution.

C:

When addressing gathering migration files without executing code, filename enumeration establishes the migration set without executing anything; the present task instead calls for choosing direct file inspection over shell execution.

D:

Searching content narrows candidates; reading the relevant paths supplies surrounding context; moving from symbol search to targeted file inspection and choosing direct file inspection over shell execution require different controls in this situation.

E:

Glob resolves matching file paths by name or path shape; the unsatisfied requirement is choosing direct file inspection over shell execution, not enumerating files by filename pattern.

Question 14

A months-long support session runs out of context after retaining every old tool payload. What strategy preserves useful continuity?

  1. Preserve approval identity, timestamp and supporting policy evidence
  2. Retain pending approvals, blockers and next required actions explicitly
  3. Compact earlier turns into a verified summary that keeps decisions and open actions
  4. Capture policy version and recheck rules that govern pending decisions

Explanation

Correct Answer

C:

Context compaction reduces token load while retaining essential state. The summary should preserve identifiers, unresolved questions and authoritative evidence rather than every verbose tool response.

Incorrect Answers

A:

High-impact decisions require an audit trail, not just a terse conclusion; preserving authorization evidence through summarization and retaining critical state while compacting history are distinct responsibilities under these constraints.

B:

Pending workflow state is essential for safe resumption; the action supports keeping unresolved control gates in condensed context but falls short of retaining critical state while compacting history in this case.

D:

Summaries are snapshots rather than proofs of current policy; that addresses detecting stale policy within condensed state, whereas this case calls for retaining critical state while compacting history.

Question 15

A research report requires both factual accuracy and coherent prose. What multi-pass process addresses these distinct concerns?

  1. Introduce independent evidence checks rather than relying solely on answer agreement
  2. Verify claims against sources before a separate editorial clarity pass
  3. Order findings by impact and evidence strength before producing recommendations
  4. Introduce a second independent validation pass against the source document

Explanation

Correct Answer

B:

Fact checking and presentation optimization use different evidence and criteria. Separating them avoids treating fluent style as proof of truth.

Incorrect Answers

A:

Because several models may share the same misconception, the measure supports avoiding false confidence from reviewer agreement rather than the needed separating factual checking from writing polish.

C:

Review output should prioritize well-supported important issues; supporting prioritizing findings after independent review cannot replace the required separating factual checking from writing polish.

D:

A dedicated verification step can compare amounts and field provenance before acceptance; this contributes to trading review latency for financial precision without meeting the need for separating factual checking from writing polish.

Question 16

A team submits thousands of malformed requests in a batch and all fail validation. What preflight practice would prevent it?

  1. Identify expiration and resubmit eligible unfinished requests separately
  2. Choose a synchronous interactive request for the latency-sensitive turn
  3. Save outcomes by input ID and create a retry queue from non-success statuses
  4. Favor asynchronous batching and budget for deferred results
  5. Test one representative request shape synchronously before scaling the batch
  6. Limit batch result access and enforce the organization's data retention policy

Explanation

Correct Answer

E:

A small dry run catches unsupported parameters and schema errors early. Large-scale retries of an invalid payload multiply wasted effort.

Incorrect Answers

A:

Deferred processing has time limits and can return expired outcomes; a method suited to recovering expired batch requests does not meet the need for prevalidating request shape before bulk submission.

B:

For keeping interactive requests out of deferred batches, a batch job runs asynchronously and completion is not immediate; however, this requirement concerns prevalidating request shape before bulk submission.

C:

Structured reconciliation lets teams preserve successful results and focus intervention on failures; that may serve constructing targeted post-batch retries, but it is no substitute for prevalidating request shape before bulk submission.

D:

Batch processing can reduce unit cost at the expense of immediate completion; this may support trading response latency for bulk-processing cost but cannot alone meet the requirement for prevalidating request shape before bulk submission.

F:

Bulk result files can concentrate sensitive information; a method suited to protecting sensitive bulk-processing results does not meet the need for prevalidating request shape before bulk submission.

Question 17

A structured result passes all type checks but violates an invoice's tax rules. What validation approach is missing?

  1. Compare extracted items against their source spans or document evidence
  2. Apply domain-specific calculation and policy rules after schema parsing
  3. Implement a bounded retry budget and move the document to an exception queue
  4. Repeat with targeted feedback identifying the invalid field and expected format

Explanation

Correct Answer

B:

Schema guarantees shape, not compliance with business calculations. Explicit rule checks can reject a typed result with incorrect tax logic.

Incorrect Answers

A:

Per-field provenance validation detects additions not supported by the original content; the deciding condition is validating business invariants after schema checks, whereas this action concerns matching extracted entities to source evidence.

C:

Repeated retries cost money and may not improve fidelity; this addresses bounding extraction repair attempts, while the question is about validating business invariants after schema checks.

D:

Feedback-based repair gives the model a specific correctable defect; that relates to targeting a retry at a correctable parse defect, not the necessary validating business invariants after schema checks.

Question 18

An extraction task has rare documents with no total amount. What example would teach correct handling?

  1. Focus on demonstrating zero, one and multiple valid extraction matches as appropriate
  2. Include a missing-field example using the schema's defined null behavior
  3. Introduce an example that correctly returns an unknown or escalation category
  4. Preserve a small representative few-shot set and evaluate the trade-off empirically

Explanation

Correct Answer

B:

A deliberate absent-value demonstration discourages invented totals. Examples should show how the parser represents unavailable information rather than silently filling it.

Incorrect Answers

A:

When addressing calibrating extraction cardinality, cardinality examples teach when to return arrays versus absent values; the present task instead calls for demonstrating missing-field semantics.

C:

Boundary examples teach that abstention is acceptable; the unsatisfied requirement is demonstrating missing-field semantics, not calibrating abstention with a boundary example.

D:

Large demonstrations consume context and tokens; in this scenario, controlling context cost from examples is secondary to demonstrating missing-field semantics.

Question 19

A local MCP tool server reads standard input and writes protocol messages to standard output. Which transport model is this?

  1. Secure the transport with TLS and authenticate clients before sensitive tool calls
  2. Inspect and approve the project server before granting it access
  3. Start the MCP server as a local stdio process
  4. Verify tool exposure, permission filters and server-side authorization

Explanation

Correct Answer

C:

Stdio transport connects the client to a child process using local streams. It differs from a network-hosted HTTP endpoint and has distinct process lifecycle concerns.

Incorrect Answers

A:

Tool inputs and results may contain confidential operational data; this can serve protecting an MCP network connection, but leaves the requirement for identifying local stdio MCP transport unmet.

B:

Repository-supplied configurations can execute processes or contact services; that addresses reviewing untrusted project MCP definitions, whereas this case calls for identifying local stdio MCP transport.

D:

Connectivity alone does not grant every capability; diagnosing a connected but unavailable MCP tool and identifying local stdio MCP transport are distinct responsibilities under these constraints.

Question 20

An extraction prompt requests 'all details' from invoices, producing irrelevant fields. What redesign reduces drift?

  1. Describe evaluation priorities such as latency, correctness, budget and operational ownership
  2. Enumerate the exact fields, types and normalization rules required downstream
  3. Mandate verbatim preservation for named entities while normalizing only designated fields
  4. Require source-backed claims and a labeled unknowns or uncertainty section
  5. Specify redaction rules and permitted data fields in the expected summary

Explanation

Correct Answer

B:

Explicit extraction targets guide attention to operationally meaningful values. A vague request leaves the output schema and business semantics undefined.

Incorrect Answers

A:

Architectural choices depend on goals and trade-offs; supporting making architecture trade-off priorities explicit cannot replace the required defining bounded extraction fields.

C:

Field-level instructions distinguish transcription from interpretation; that serves preserving exact source spelling where required, whereas the case requires defining bounded extraction fields.

D:

Because requiring evidence mapping and explicit uncertainty discourages treating unverified inference as fact, the measure supports separating sourced conclusions from unknowns rather than the needed defining bounded extraction fields.

E:

Data minimization instructions make the output contract sensitive to privacy; its benefits for minimizing sensitive data in summaries do not resolve the need for defining bounded extraction fields.

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