Anthropic CCA-F Practice Test 56
Question 550
A price service fails and the app shows yesterday's cached price as current. What is the proper presentation?
- Mark the failed feed unavailable and label the report partial until requirements are met
- Preserve timeout as an explicit failure rather than an empty successful search result
- Open a circuit breaker after repeated failures and use a controlled fallback path
- Label the cached value with its timestamp and source rather than claiming it is live
Explanation
Correct Answer
D:Fallback data can be useful but has different freshness guarantees. Explicit age protects users from stale-value decisions. A cached price can be offered only with its source timestamp and staleness clearly labelled. Presenting yesterday's number as current risks an unsupported customer commitment.
Incorrect Answers
A:Declaring a feed unavailable is appropriate when no useful verified price exists, but this application explicitly has a cached price. It may use that value only if it is clearly labeled with its source and age and the user's decision can tolerate staleness; never present yesterday's price as live.
B:A timeout does not establish that no documents exist. Although useful elsewhere, it is not the direct mechanism to identify fallback data and its freshness limitations.
C:A circuit breaker prevents repeated calls to a known-failing dependency while allowing recovery probes later. This alternative does not establish how to identify fallback data and its freshness limitations in the stated workflow.
Question 551
A retrieval specialist times out but a report writer receives an empty evidence list. What result should the dependent agent receive?
- Carry partial status and named gaps into the coordinator's synthesis
- Send a specific validation failure and a bounded correction request
- Propagate the explicit timeout status and prevent a no-findings conclusion
- Retain each failure's identity and severity in the workflow error ledger
Explanation
Correct Answer
C:The relevant control is to propagate the explicit timeout status and prevent a no-findings conclusion. A missing response is not the same as a completed search showing no evidence. This matters when a retrieval specialist times out but a report writer receives an empty evidence list. A blank result is not equivalent to a tool failure, and the downstream writer must not hide that difference.
Incorrect Answers
A:Incomplete research should not be promoted into a fully verified result. The required action is “Propagate the explicit timeout status and prevent a no-findings conclusion”; the other measure addresses a separate concern. A missing response is not the same as a completed search showing no evidence.
B:Structured failures enable targeted recovery and prevent malformed downstream input. Here, a retrieval specialist times out but a report writer receives an empty evidence list. Propagate the explicit timeout status and prevent a no-findings conclusion. A missing response is not the same as a completed search showing no evidence.
D:A visible failure chain makes troubleshooting and user-facing limitations accurate. The proposed option cannot provide the specific “Propagate the explicit timeout status and prevent a no-findings conclusion” behavior that the case requires. A missing response is not the same as a completed search showing no evidence.
Question 552
A delegated analysis returns partial success and two missing sources. How should this failure be reported to the coordinator?
- Send a specific validation failure and a bounded correction request
- Retain each failure's identity and severity in the workflow error ledger
- Trace the claim to its source and request correction or mark it unsupported
- Carry partial status and named gaps into the coordinator's synthesis
Explanation
Correct Answer
D:Incomplete research should not be promoted into a fully verified result. For this situation, the system should carry partial status and named gaps into the coordinator's synthesis. The factual trigger is a delegated analysis returns partial success and two missing sources. Propagate actual error state and partial work across every handoff, preserving the originating task ID.
Incorrect Answers
A:Structured failures enable targeted recovery and prevent malformed downstream input. Here, a delegated analysis returns partial success and two missing sources. Carry partial status and named gaps into the coordinator's synthesis. Incomplete research should not be promoted into a fully verified result.
B:A visible failure chain makes troubleshooting and user-facing limitations accurate. The required action is “Carry partial status and named gaps into the coordinator's synthesis”; the other measure addresses a separate concern. Incomplete research should not be promoted into a fully verified result.
C:Error handling must preserve provenance and not fabricate a replacement source. This is not a substitute for “Carry partial status and named gaps into the coordinator's synthesis”; that control is necessary in the stated circumstances. Incomplete research should not be promoted into a fully verified result.
Question 553
A subagent reports an invalid schema in its output. Which error-propagation behavior preserves the correct meaning?
- Send a specific validation failure and a bounded correction request
- Retain each failure's identity and severity in the workflow error ledger
- Trace the claim to its source and request correction or mark it unsupported
- Propagate the explicit timeout status and prevent a no-findings conclusion
Explanation
Correct Answer
A:When a subagent reports an invalid schema in its output, the system should send a specific validation failure and a bounded correction request. Structured failures enable targeted recovery and prevent malformed downstream input. A blank result is not equivalent to a tool failure, and the downstream writer must not hide that difference.
Incorrect Answers
B:A visible failure chain makes troubleshooting and user-facing limitations accurate. The direct remedy is “Send a specific validation failure and a bounded correction request”, which the alternative does not itself accomplish. Structured failures enable targeted recovery and prevent malformed downstream input.
C:Error handling must preserve provenance and not fabricate a replacement source. Here, a subagent reports an invalid schema in its output. Send a specific validation failure and a bounded correction request. Structured failures enable targeted recovery and prevent malformed downstream input.
D:A missing response is not the same as a completed search showing no evidence. This is not a substitute for “Send a specific validation failure and a bounded correction request”; that control is necessary in the stated circumstances. Structured failures enable targeted recovery and prevent malformed downstream input.
Question 554
Several tool failures are swallowed by an orchestration wrapper. What result should the dependent agent receive?
- Trace the claim to its source and request correction or mark it unsupported
- Retain each failure's identity and severity in the workflow error ledger
- Propagate the explicit timeout status and prevent a no-findings conclusion
- Carry partial status and named gaps into the coordinator's synthesis
Explanation
Correct Answer
B:Several tool failures are swallowed by an orchestration wrapper. The relevant control is to retain each failure's identity and severity in the workflow error ledger. A visible failure chain makes troubleshooting and user-facing limitations accurate. Propagate actual error state and partial work across every handoff, preserving the originating task ID.
Incorrect Answers
A:Error handling must preserve provenance and not fabricate a replacement source. Here, several tool failures are swallowed by an orchestration wrapper. Retain each failure's identity and severity in the workflow error ledger. A visible failure chain makes troubleshooting and user-facing limitations accurate.
C:A missing response is not the same as a completed search showing no evidence. The direct remedy is “Retain each failure's identity and severity in the workflow error ledger”, which the alternative does not itself accomplish. A visible failure chain makes troubleshooting and user-facing limitations accurate.
D:Incomplete research should not be promoted into a fully verified result. The required action is “Retain each failure's identity and severity in the workflow error ledger”; the other measure addresses a separate concern. A visible failure chain makes troubleshooting and user-facing limitations accurate.
Question 555
A downstream reviewer sees an unsupported citation from a prior handoff. How should this failure be reported to the coordinator?
- Propagate the explicit timeout status and prevent a no-findings conclusion
- Carry partial status and named gaps into the coordinator's synthesis
- Trace the claim to its source and request correction or mark it unsupported
- Send a specific validation failure and a bounded correction request
Explanation
Correct Answer
C:The relevant control is to trace the claim to its source and request correction or mark it unsupported. Error handling must preserve provenance and not fabricate a replacement source. This matters when a downstream reviewer sees an unsupported citation from a prior handoff. A blank result is not equivalent to a tool failure, and the downstream writer must not hide that difference.
Incorrect Answers
A:A missing response is not the same as a completed search showing no evidence. Here, a downstream reviewer sees an unsupported citation from a prior handoff. Trace the claim to its source and request correction or mark it unsupported. Error handling must preserve provenance and not fabricate a replacement source.
B:Incomplete research should not be promoted into a fully verified result. Here, a downstream reviewer sees an unsupported citation from a prior handoff. Trace the claim to its source and request correction or mark it unsupported. Error handling must preserve provenance and not fabricate a replacement source.
D:Structured failures enable targeted recovery and prevent malformed downstream input. Here, a downstream reviewer sees an unsupported citation from a prior handoff. Trace the claim to its source and request correction or mark it unsupported. Error handling must preserve provenance and not fabricate a replacement source.
Question 556
A coding agent is asked to modify checkout behavior but cannot locate the main entry point. What should it do first?
- Inspect environment-specific configuration layers and precedence at runtime
- Inspect dependency manifests and lockfile resolution across packages
- Find all importers and consumers of the changed type and test compatibility
- Search route and handler names with Grep and inspect the relevant files
Explanation
Correct Answer
D:Targeted symbol discovery narrows the codebase efficiently. Blindly rewriting likely filenames risks modifying the wrong path. Use Grep to find likely route and handler identifiers, then Read their surrounding definitions to trace the checkout entry point. Opening every repository file would waste context and may hide the relevant dependency.
Incorrect Answers
A:Environment-specific configuration precedence matters when a program works differently in CI, but this agent has not located the checkout code at all. Begin by searching the repository for route names, handler symbols or checkout references, then inspect the identified files before investigating runtime configuration.
B:Conflicting transitive versions can make apparently identical types incompatible. This is not a substitute for “Search route and handler names with Grep and inspect the relevant files”; that control is necessary in the stated circumstances. Use Grep to find likely route and handler identifiers, then Read their surrounding definitions to trace the checkout entry point.
C:A shared interface has effects beyond its definition. That addresses a different part of the system, not the requirement to locate actual execution entry points before editing.
Question 557
A large repository contains many services and the agent needs to find configuration manifests. Which exploration method fits?
- Use filename-pattern discovery and then inspect only relevant manifests
- Trace active build flags and inspect the platform-specific implementation selected at runtime
- Find all importers and consumers of the changed type and test compatibility
- Inspect dependency manifests and lockfile resolution across packages
Explanation
Correct Answer
A:Structural search answers where configuration lives without dumping irrelevant file content into context. Start with Glob or an equivalent filename search to find configuration paths, then open relevant manifests with Read. This reduces token use and avoids unrelated files while preserving enough context for a safe change.
Incorrect Answers
B:Repository file names alone do not identify which conditional branch the build uses. It would leave the stated need to map relevant codebase paths efficiently unresolved.
C:A shared interface has effects beyond its definition. The decisive requirement here is to map relevant codebase paths efficiently. Checking all type consumers is important after a known type change, but the immediate problem is locating the code across many services. First use targeted repository search.
D:Conflicting transitive versions can make apparently identical types incompatible. This alternative does not establish how to map relevant codebase paths efficiently in the stated workflow.
Question 558
An API response field is wrong but the endpoint handler appears correct. What investigation should follow?
- Inspect environment-specific configuration layers and precedence at runtime
- Trace the field through its service and data-access call chain
- Search for dynamic references, generated clients and external contracts as well as static imports
- Locate the generator source schema and regeneration procedure first
Explanation
Correct Answer
B:Symptoms may originate below the visible handler. Following references and return transformations identifies the true fault location. Trace the incorrect field through handler, service logic, adapter and persistence layers. The outward API handler can be correct while an upstream mapper or query supplies the wrong value.
Incorrect Answers
A:The same code can behave differently when configuration sources differ. Although useful elsewhere, it is not the direct mechanism to trace behavior across the relevant call graph.
C:Not all dependencies appear as direct code imports. This alternative—Search for dynamic references, generated clients and external contracts as well as static imports—does not implement the specific operation that the question asks about. The required control is Trace the field through its service and data-access call chain. Symptoms may originate below the visible handler.
D:Generated files are derived artifacts. The scenario instead requires a way to trace behavior across the relevant call graph. Generated client templates are irrelevant if the observed field value was modified by a database mapper or service layer. Trace the live field through the actual request path.
Question 559
A data type changes in a shared package and dependent services fail. What should the agent inspect?
- Inspect dependency manifests and lockfile resolution across packages
- Find all importers and consumers of the changed type and test compatibility
- Locate the generator source schema and regeneration procedure first
- Use filename-pattern discovery and then inspect only relevant manifests
Explanation
Correct Answer
B:A public function name can be referenced by direct imports, generated client code, runtime dispatch and external integration contracts. Searching all known usages before a rename identifies affected consumers and tests that must be updated together. A local declaration-only edit can appear correct while leaving runtime paths or published API callers broken.
Incorrect Answers
A:Conflicting transitive versions can make apparently identical types incompatible. That addresses a different part of the system, not the requirement to check downstream dependencies before shipping interface changes.
C:Generated files are derived artifacts. This capability does not directly check downstream dependencies before shipping interface changes. Regenerating client code cannot resolve a type incompatibility in the shared package until all consuming imports and version selections are identified. Inventory downstream dependencies.
D:Structural search answers where configuration lives without dumping irrelevant file content into context. The decisive requirement here is to check downstream dependencies before shipping interface changes.
Question 560
A defect appears only in the Windows build because the repository selects different parser implementations by build flag. Which exploration step identifies the code actually executed?
- Use the failing tests to narrow affected modules before changing unrelated code
- Trace active build flags and inspect the platform-specific implementation selected at runtime
- Inspect environment-specific configuration layers and precedence at runtime
- Use filename-pattern discovery and then inspect only relevant manifests
Explanation
Correct Answer
B:Repository file names alone do not identify which conditional branch the build uses. Checking configuration values and call paths avoids repairing an inactive implementation. Conditional builds can select different parsers for Windows and other targets. Inspect the relevant build flags and active implementation before altering a generic file that is not compiled on the failing platform.
Incorrect Answers
A:Using test failures to narrow affected modules is useful after observing a failing test, but here the defect's defining clue is platform-specific parser selection. Trace build flags, package conditions and the Windows implementation actually loaded; otherwise generic test triage may overlook the alternate parser.
C:Configuration precedence may explain some CI differences, but the stated Windows-only defect follows conditional selection of parser implementations. Inspect the active platform/build flags and the corresponding code path before changing generic environment values. The failure is an implementation branch, not necessarily a configuration override.
D:Structural search answers where configuration lives without dumping irrelevant file content into context. This alternative does not establish how to trace the active platform-specific execution path in the stated workflow.