Claude Certified Developer – Foundations (CCDV-F) practice exam and questions

Decide on stop_reason instead: run the requested tools and call again while it is tool_use, and finish on any other value.

Raise max_tokens on each request so Claude has room to finish the lookup and write the full answer in one reply.

Set tool_choice to any on each request so Claude must call a tool in its turn instead of writing a text reply.

Add a system prompt rule telling Claude to write no text in a turn where it calls a tool, so each reply holds just the tool calls.

Send each run's tickets as concurrent Messages API requests, so the tags return while that run's job is still working.

Submit each run as a message batch, since test batches of this size came back well inside the 10-minute window.

Run the tagging once a night as one large message batch, and show the previous day's sentiment on the dashboard.

Submit each run as a message batch with a cache breakpoint after the shared instructions, so each batch processes faster.

Allow null for invoice_date in the schema, and send invoices whose date comes back null to the manual review queue.

Keep invoice_date required in the schema, and add a system prompt rule telling Claude not to guess a missing date.

Add a pattern to invoice_date in the schema so the field accepts only strings in the YYYY-MM-DD date format.

Raise the effort setting so that Claude examines each scanned page more carefully before it fills in invoice_date.

Store the key in the phone's secure keystore on first launch, fetching it once from the team's remote configuration service.

Route requests through a serverless function that adds the API key, so the app binary holds just the function's public URL.

Send the app's requests to the team's backend, which verifies each user's sign-in and calls the Messages API with its key.

Give the app its own key in a separate workspace with a low spend limit, so an extracted copy can do little damage.

Enable prompt caching on the conversation so the growing list of earlier tool results is read from the cache on later turns.

Enable programmatic tool calling so Claude writes code in a code execution container that calls the tool and sums the results.

Turn on context editing so older tool results are cleared from the conversation once later SKUs have been fetched and counted.

Let Claude request many get_stock calls in parallel in each turn so that far fewer model turns are needed to fetch every record.