Module quiz

21 questions · pass bar 80% · retry as often as you like — your best score counts toward the gate.

  1. 1.Why must you resend the full message history on every API call?
  2. 2.In one tool-call round trip with the Anthropic API, what is the correct message sequence after the model returns stop_reason: "tool_use"?
  3. 3.A 10-turn conversation averages 500 tokens per turn. Roughly how many input tokens does the API call at turn 10 consume, and why?
  4. 4.What is the difference between tool calling and structured output, and when should you use a JSON-schema output instead of a tool?
  5. 5.The model returns malformed JSON for a tool call. What's the right first mitigation?
  6. 6.What does temperature actually do, and what happens if you send temperature=0.7 to a mid-2026 frontier Claude model like claude-sonnet-5?
  7. 7.How do API rate limits typically work, and what's a correct backoff implementation?
  8. 8.Which error should you NEVER automatically retry?
  9. 9.What is prompt caching and when does it cut agent costs dramatically?
  10. 10.Why should tool descriptions be written as carefully as prompts?
  11. 11.What happens if you send a tool_result whose tool_use_id doesn't match a tool call from the immediately preceding assistant message?
  12. 12.Your streaming agent shows nothing for 8 seconds, then dumps the full answer. What's the most likely bug?
  13. 13.What are adaptive thinking and the effort parameter on current frontier models?
  14. 14.Your pipeline classifies one million support tickets per day and drafts personalized responses for the ~2% that escalate. What's the cost-sane model strategy?
  15. 15.What is the most reliable way to get schema-conforming JSON from a current Anthropic model?
  16. 16.Your agent consults the same 300-page PDF manual on every session. What's the right way to send it, and what else do you need for it to be economical?
  17. 17.A nightly job classifies 800K documents; results are needed by 8 a.m. What's the right API surface, and what do you gain?
  18. 18.While streaming a response, your code calls json.loads() on each tool-argument delta as it arrives, and crashes. What's actually on the wire, and what's the fix?
  19. 19.Inside the agent loop, a tool raises an exception (file not found). What does production-grade code do?
  20. 20.Which set of fields belongs in the per-call structured log line of a production LLM service?
  21. 21.Your traffic is bursty: a flurry of requests every 20 minutes, silence in between. Default 5-minute-TTL caching shows writes but almost no reads. What's the calculus for switching to the 1-hour TTL?