Claude SDK & Anthropic APIs: free practice questions

The Batch API is designed for high-volume, latency-insensitive workloads where results can be collected asynchronously. Option A (200,000 tickets overnight) and Option C (50,000 product descriptions weekly) are both classic batch workloads—large volume, no real-time requirement, and significant cost savings (Batch API is typically 50% cheaper). Option B is wrong because a live chatbot with 2-second SLA requires synchronous, low-latency calls—the Batch API is asynchronous with processing times that can take minutes to hours. Option D is wrong because streaming to an IDE plugin is inherently interactive and real-time. Option E is wrong because real-time search re-ranking requires synchronous, sub-second responses.

Options B and D are correct. B is correct: the `ephemeral` cache TTL is a minimum of 5 minutes, and cache hits refresh (extend) the TTL — so actively used cache entries remain valid longer. D is correct: reading from cache costs significantly fewer tokens than a full cache miss (cache reads are billed at a lower rate than standard input tokens), making prompt caching cost-effective for repeated prefixes like long system prompts. Option A is wrong because it states the TTL is 'fixed at 5 minutes' and implies no refreshing — the TTL resets on hits. Option C is wrong; cache entries do expire after inactivity and are not permanent. Option E is wrong; cache entries are scoped to the API key that created them and are not shared across keys even within the same organization. Option F is wrong; while `ephemeral` is the primary type, Anthropic's documentation references other cache behaviors and the landscape may evolve — more importantly, stating it as the only type is misleading.

Option A is correct: the Claude Code SDK supports scoping subagent capabilities by specifying an `allowed_tools` list. By explicitly listing only file-writing tools, the developer ensures the subagent cannot invoke shell execution or other unauthorized tools, implementing the principle of least privilege. Option B is incorrect: `max_tokens` controls output token length, not the number of tool calls or action steps; a subagent could still execute a shell command within a short response. Option C is incorrect: `tool_choice: {"type": "none"}` disables all tool use, which would prevent the file-writing task from being completed, not just shell execution. Option D is incorrect: the Claude Code SDK does not raise a `PermissionError` automatically based on tool type; permission scoping must be configured explicitly by the developer.

Setting `tool_choice` to `{"type": "tool", "name": "log_interaction"}` forces Claude to call exactly the named tool on every turn. This is the correct way to mandate use of a specific tool. Option B (`type: required`) forces Claude to use *some* tool but does not specify which one—Claude could choose any available tool. Option C (`type: auto`) is the default and allows Claude to decide whether and which tool to use, so it does not guarantee `log_interaction` is called. Option D uses a non-existent `"any"` type combined with `name`; `"any"` is not a valid `tool_choice` type in the Anthropic API.

The Anthropic tool `input_schema` follows JSON Schema conventions. Required parameters are listed in a top-level `required` array, while optional parameters are simply absent from that array. Option A correctly places only `query` in `required` and omits `max_results` from it, while also including a `default` hint. Option B incorrectly embeds `required` as a per-property boolean attribute, which is not valid JSON Schema syntax used by the Anthropic API. Option C incorrectly lists both fields as required, which would force Claude to always provide `max_results`. Option D uses `"type": "function"` at the schema root and a `parameters` key—this is OpenAI function-calling format, not the Anthropic `input_schema` format.