最新的Anthropic Claude Certified Developer-Foundations - CCDV-F免費考試真題

A. Define a clear schema, structure the prompt to request output in that schema, and validate Claude's output against the schema before passing it downstream.

B. Structure the prompt to request output in a schema that is described in plain English, with downstream systems parsing whatever shape Claude produces.

C. Define a clear schema and structure the prompt to request output in that schema, with downstream systems handling any validation needed.

D. Avoid structured output and use free-form text everywhere instead, on the grounds that free-form text is more flexible and handles edge cases better than structured schemas.

A. Keeping the deprecated parameter in place while writing a wrapper function around it to insulate the rest of the codebase from the eventual change.

B. Swapping all 40 call sites in a single change right away to prevent drawn-out migration work that will delay ongoing functioning.

C. Adding regression tests to cover the parameter's behavior, then migrating call sites in batches that you validate against regression tests.

D. Migrating all 40 call sites in a single change near the removal date to ensure the deprecated parameter continues to work as long as possible.

A. Confirm that the variation is a bug in the application's request construction and start investigating which part of the application is producing the inconsistent prompt text.

B. Tell the teammate that the variation depends on which Claude model serves the request, and recommend pinning the application to a single model version to make the output behavior consistent.

C. Tell the teammate that the variation is caused by the application sending two different prompts on different runs and propose a code change that pins the system prompt to a single version.

D. Explain that LLMs generate output token by token, and variation in how prompt content appears in output is a property of generation, not a bug in request construction.

A. Switch every feature to the smallest model to cut cost broadly across the application during normal operation.

B. Reduce token usage uniformly across all features by half, applying the cut evenly across the application during normal operation.

C. Add token usage tracking by feature to the application's logging so the team can identify which features drive cost before recommending changes.

D. Tell the team that cost growth is unavoidable as the application scales and that no investigation will change the trajectory of the application's cost over the next several quarters.

A. Use multi-shot prompting by including the labeled examples in the prompt so Claude can match the desired output format on each request.

B. Use zero-shot prompting and rely on the model's general capability to produce the desired output format consistently across all incoming requests.

C. Train a custom model on the labeled examples before deployment so that the application does not need to include the examples in any prompt during operation.

D. Embed the examples in a database and retrieve them at runtime as reference material for the team.

A. Truncate the conversation so the model sees only the most recent turn during each subsequent response.

B. Increase the context window size so all turns of the conversation remain visible to the model in full detail.

C. Reset the conversation after every turn so the model loses all prior turns when generating a response.

D. Apply compaction to summarize older portions of the conversation so the gist remains while the specifics carry less weight.

A. Zero-shot prompting alone with no model option adjustments, which keeps the application's configuration as simple as possible.

B. Fast mode, which prioritizes the lowest possible latency at the expense of reasoning depth on complex tasks.

C. A smaller model with a tighter context window, which encourages the model to focus its limited capacity on the task.

D. Extended thinking, which lets the model reason through the differential diagnosis steps before producing the final recommendation.

A. A smaller, faster Claude model, because the task is straightforward and the workload prioritizes latency and per-request cost at scale.

B. Multiple Claude models in series, where each request runs through more than one model and the application combines the outputs into a final classification.

C. A mid-tier Claude model selected by default, because mid-tier models balance quality and cost in a way the team can apply across most tasks.

D. The largest, highest-capability Claude model, to maximize quality on every classification the application produces during normal operation across all requests.

A. Add Claude hooks that intercept database operations and apply deterministic checks, such as blocking destructive queries or requiring approval, before the queries execute.

B. Add a system prompt instruction telling the agent to be careful with database operations on every request the application handles during normal operation across all incoming traffic.

C. Switch to a higher-capability Claude model on the grounds that a more capable model is less likely to run destructive queries during normal operation across all requests.

D. Run the agent only during business hours when humans are available to monitor its activity, treating the schedule as the primary control mechanism for destructive operations.

A. A workflow for the standard path with an agent invoked at the decision point for unexpected formats.

B. A fully autonomous agent that handles every invoice from start to finish across all formats.

C. A single large prompt that processes every incoming invoice, both standard and unexpected, in one model call.

D. A manager agent that delegates each step of standard invoice processing to a dedicated subagent, with a separate subagent handling each unexpected format.