Any2Quiz
Certifications / Claude Architect Foundations

Tool Design & MCP Integration Quiz - Certifications Claude Architect Foundations

Free practice quiz: Tool Design & MCP Integration (Certifications Claude Architect Foundations). 62 multiple-choice questions with answers and explanations. Play online, no sign-up needed.

Play all 62 questions free

Sample questions

  1. What is the most likely root cause of the extra get_customer call?

    1. The alphabetical tool order leads the model to pick the first plausible match.
    2. The simpler input schema leads the model to prefer get_customer over lookup_order.
    3. The system prompt instruction to verify customers creates a keyword association.
    4. The overlapping tool descriptions leave the model unable to distinguish the tools.
    Show answer

    Answer: D. The overlapping tool descriptions leave the model unable to distinguish the tools.

    Tool descriptions are the primary mechanism the model uses for tool selection. When two tools have near-identical descriptions, the model cannot tell when each should be used and may pick the wrong one. The fix is to clearly differentiate their purpose, inputs, outputs, and boundaries.

  2. What structured response design should the MCP tool use to let the agent make appropriate retry decisions?

    1. Return errorCategory transient with isRetryable true for both access denied and no entries found so the agent retries both cases.
    2. Return errorCategory permission with isRetryable false for access denied, and isError false with an empty result for a valid no-match query.
    3. Return a generic error string for both cases and let the agent infer from the specific file name alone whether to retry.
    4. Return errorCategory validation with isRetryable false for both access denied and no entries found so the agent never retries.
    Show answer

    Answer: B. Return errorCategory permission with isRetryable false for access denied, and isError false with an empty result for a valid no-match query.

    Permission failures are non-retryable access failures, while an empty result with isError false represents a successful query with no matches. Structured metadata prevents both wasted retries and misinterpretation of valid empty results. Uniform or misclassified responses remove the distinction the agent needs.

  3. What configuration change is the best way to guarantee the first turn calls get_customer?

    1. Set tool_choice to {"type": "tool", "name": "get_customer"} for the first assistant turn.
    2. Add a system prompt that says, "Always call get_customer before responding to the customer."
    3. Remove lookup_order, process_refund, and escalate_to_human so get_customer is the only tool the agent can call.
    4. Set tool_choice: "any" on all turns to guarantee the agent calls a tool first.
    Show answer

    Answer: A. Set tool_choice to {"type": "tool", "name": "get_customer"} for the first assistant turn.

    Forced tool selection names get_customer as the first-turn tool, making the call deterministic. Prompting only steers behavior, tool_choice any does not select the right tool, and removing other tools prevents later support workflows.

  4. Which setup should the team commit to the repository?

    1. Commit a project .mcp.json setting Authorization to Bearer ${GITHUB_TOKEN} for the GitHub HTTP server.
    2. Commit a project .mcp.json embedding the shared GitHub token directly in its Authorization header.
    3. Commit a project .mcp.json running npx github-mcp-server and passing the token as a CLI argument.
    4. Create a user-level ~/.claude.json for each developer with a personal GitHub token and no project file.
    Show answer

    Answer: A. Commit a project .mcp.json setting Authorization to Bearer ${GITHUB_TOKEN} for the GitHub HTTP server.

    Project-scoped .mcp.json is the correct shared-team configuration, and ${GITHUB_TOKEN} environment variable expansion keeps credentials out of the repository. Embedding tokens or passing them on the command line leaks secrets, while user-level config would not be shared with the team.

  5. What is the best change to make to the Grep invocation?

    1. Set glob to '**/*.py' to limit the search scope to only Python source files.
    2. Set output mode to content to show matches with file names and line numbers.
    3. Set multiline to true to return the full text of each matching line from Grep.
    4. Set type to 'py' to receive richer language-aware results from Grep output.
    Show answer

    Answer: B. Set output mode to content to show matches with file names and line numbers.

    Grep defaults to files_with_matches, which returns only file paths. Switching to content mode is the documented way to get matching lines with file and line numbers. Glob and type narrow scope but do not add line content; multiline changes how patterns match, not the output fields.

Take the full quiz

More quizzes

Browse all Claude Architect Foundations quizzes