Free Claude Certified Architect – Foundations Practice Questions

Sample Practice Questions

Try these 10 sample questions from the Claude Certified Architect – Foundations practice bank. Each includes the answer and explanation. Continue with the interactive practice above to review the full question set.

1. 1A client-executed tool has finished and a Claude application is preparing its next Messages API request. Which message carries the result back to Claude?

   1. A.A user message containing a tool_result blockCorrect answer: A
   2. B.An assistant message containing a tool_result block
   3. C.A user message containing a new tool_use block
   4. D.An assistant message containing a new tool_use block

   Explanation: The documentation says, “Continue the conversation by sending a new message with the role of user, and a content block containing the tool_result type”. The application sends the result in a user message; an assistant message is the wrong role. A tool_use block denotes a tool request, not the returned tool result, so neither option using tool_use supplies the result.

   Reference: Claude Platform Docs: Handle tool calls — Format results and continue the conversation

2. 2A tool accepts customer identifiers in the form C- followed by three digits; order identifiers use O- followed by three digits. Which tool-description example clarifies the customer input format?

   1. A.customer_id: C-123Correct answer: A
   2. B.customer_id: O-123
   3. C.order_id: C-123
   4. D.order_id: O-123

   Explanation: The guide highlights “input formats, example queries, edge cases, and boundary explanations” in tool descriptions. Under the stated conventions, customer_id: C-123 is the customer example; the other examples use the order field or the wrong identifier prefix.

   Reference: Claude Certified Architect – Foundations Exam Guide — page 9, Task Statement 2.1

3. 3A developer puts a team coding instruction in ~/.claude/CLAUDE.md and expects teammates to receive it when they pull the repository. What scope issue does the exam guide identify?

   1. A.User-level instructions apply to that user and are not shared through project version controlCorrect answer: A
   2. B.User-level instructions apply only to files matching project rule paths
   3. C.User-level instructions are shared through the project's .mcp.json
   4. D.User-level instructions are loaded only by a project slash command

   Explanation: The guide states “user-level settings apply only to that user” and are “not shared with teammates via version control.” The other choices confuse user instructions with project rule paths, MCP server configuration, or slash commands.

   Reference: Claude Certified Architect – Foundations Exam Guide — page 12, Task Statement 3.1

4. 4A code-review prompt generates too many false positives for harmless style differences. Which change does the guide recommend for improving precision?

   1. A.Ask the model only to report high-confidence findings
   2. B.Increase the number of review passes without defining categories
   3. C.Define specific categories of issues to report and skipCorrect answer: C
   4. D.Ask the model only to be generally conservative

   Explanation: The guide recommends “specific review criteria that define which issues to report ... versus skip” rather than relying on confidence-based filtering. Vague conservatism or a confidence instruction does not define the boundary; extra passes without categories do not supply that boundary either.

   Reference: Claude Certified Architect – Foundations Exam Guide — page 16, Task Statement 4.1

5. 5A support conversation may be summarized repeatedly. The agent needs the exact order number, amount, and status in later prompts. Which arrangement does the guide recommend?

   1. A.Keep those facts in a persistent case-facts block outside summarized historyCorrect answer: A
   2. B.Keep those facts only in the evolving conversation summary
   3. C.Keep those facts only in the original tool-result message
   4. D.Keep those facts only in a subagent's isolated conversation

   Explanation: The guide recommends a persistent “case facts” block containing transactional facts outside summarized history. An evolving summary, old tool output, or isolated subagent conversation does not ensure those facts remain in subsequent prompts.

   Reference: Claude Certified Architect – Foundations Exam Guide — page 20, Task Statement 5.1

6. 6A Claude response requests several client-executed tool calls. What should the application put in each returned tool_result block to associate it with the correct request?

   1. A.The next assistant message's id in tool_use_id
   2. B.The requested tool's name in tool_use_id
   3. C.The tool-use request's id in tool_use_idCorrect answer: C
   4. D.The user message's id in tool_use_id

   Explanation: The documentation defines “tool_use_id: The id of the tool use request this is a result for.” A tool name identifies a tool rather than the specific invocation, while the user-message and next-assistant-message IDs do not identify the tool-use request whose result is being returned.

   Reference: Claude Platform Docs: Handle tool calls — Format results, tool_use_id

7. 7A generic analyze_document tool combines extraction, summarization, and claim verification. The team wants three tools with separate contracts and no combined generic operation. Which tool set matches that design?

   1. A.extract_data_points, summarize_content, verify_claim_against_sourceCorrect answer: A
   2. B.extract_data_points, summarize_content, analyze_document
   3. C.extract_data_points, analyze_document, verify_claim_against_source
   4. D.analyze_document, summarize_content, verify_claim_against_source

   Explanation: The guide illustrates splitting a generic tool into “extract_data_points, summarize_content, and verify_claim_against_source.” Each replacement describes one operation; retaining the broad analyze_document in any position leaves an overlapping generic boundary.

   Reference: Claude Certified Architect – Foundations Exam Guide — page 9, Task Statement 2.1

8. 8A project wants its CLAUDE.md to import the file docs/api-standards.md. Which reference uses the documented import syntax?

   1. A.A plain @docs/api-standards.md referenceCorrect answer: A
   2. B.A quoted "@docs/api-standards.md" reference
   3. C.A Markdown link to docs/api-standards.md
   4. D.A Markdown code span containing @docs/api-standards.md

   Explanation: The documentation specifies “@path/to/import syntax,” says quoted paths are not imported, and skips Markdown code spans. An ordinary Markdown link is a reference for readers rather than the documented @path import.

   Reference: Claude Code documentation: Memory — Import additional files

9. 9A team's review rubric assigns high severity to authorization bypasses and low severity to naming-only issues. Which example pair applies those criteria consistently?

   1. A.Authorization bypass → high; naming-only issue → lowCorrect answer: A
   2. B.Authorization bypass → low; naming-only issue → high
   3. C.Authorization bypass → high; naming-only issue → high
   4. D.Authorization bypass → low; naming-only issue → low

   Explanation: The guide recommends “explicit severity criteria with concrete code examples for each severity level.” Applying the stated rubric assigns the authorization bypass high and the naming-only issue low; the other pairs misclassify at least one example.

   Reference: Claude Certified Architect – Foundations Exam Guide — page 16, Task Statement 4.1

10. 10An order lookup returns order ID, status, amount, shipping carrier, and warehouse ID. For a refund question, only order ID, status, and amount are relevant. Which retained field set follows the guide's context-trimming approach?

    1. A.Order ID, status, and amountCorrect answer: A
    2. B.Order ID, status, and shipping carrier
    3. C.Order ID, amount, and warehouse ID
    4. D.Status, shipping carrier, and warehouse ID

    Explanation: The guide recommends “Trimming verbose tool outputs to only relevant fields.” The first set retains the three fields identified as relevant; each other set replaces one or more of them with a field identified as irrelevant.

    Reference: Claude Certified Architect – Foundations Exam Guide — page 20, Task Statement 5.1