Free Anthropic CCAR-F Exam Questions
Absolute Free CCAR-F Exam Practice for Comprehensive Preparation
-
Anthropic CCAR-F Exam Questions
-
Provided By: Anthropic
- Exam: Claude Certified Architect - Foundations
- Certification: Anthropic Claude Certified Architect - Foundations
-
Total Questions: 152
-
Updated On: Aug 19, 2026
-
Rated: 4.9 |
-
Online Users: 304
-
Question 1
-
You are integrating Claude Code into your Continuous Integration/Continuous Deployment (CI/CD) pipeline.The system runs automated code reviews, generates test cases, and provides feedback on pull requests. Youneed to design prompts that provide actionable feedback and minimize false positives.Your pipeline includes a release-notes generation step that classifies and summarizes approximately 200commits at the end of each weekly release cycle. Each commit is currently sent as a separate Messages APIrequest using a Sonnet-tier Claude model. The release notes are not needed until the following morning,providing approximately 12 hours of acceptable latency.Your team must reduce the per-token API cost while retaining the same model, prompts, and output quality.Which approach satisfies all these constraints?
Answer: B
-
You are integrating Claude Code into your Continuous Integration/Continuous Deployment (CI/CD) pipeline.The system runs automated code reviews, generates test cases, and provides feedback on pull requests. Youneed to design prompts that provide actionable feedback and minimize false positives.Your pipeline includes a release-notes generation step that classifies and summarizes approximately 200commits at the end of each weekly release cycle. Each commit is currently sent as a separate Messages APIrequest using a Sonnet-tier Claude model. The release notes are not needed until the following morning,providing approximately 12 hours of acceptable latency.Your team must reduce the per-token API cost while retaining the same model, prompts, and output quality.Which approach satisfies all these constraints?
-
Question 2
-
In production, you observe that simple fact-checking queries—for example, “What year was the Paris Climate
Agreement signed?”—traverse all four subagents sequentially, consuming more than 40 seconds and
significant tokens per query. Complex comparative research benefits from the full pipeline. Your query
distribution is diverse and evolving as users discover new applications. What is the most effective approach to
optimize for varying query complexity?
Answer: C
-
In production, you observe that simple fact-checking queries—for example, “What year was the Paris Climate
Agreement signed?”—traverse all four subagents sequentially, consuming more than 40 seconds and
significant tokens per query. Complex comparative research benefits from the full pipeline. Your query
distribution is diverse and evolving as users discover new applications. What is the most effective approach to
optimize for varying query complexity?
-
Question 3
-
You are building a customer support resolution agent using the Claude Agent SDK. The agent handles highambiguity requests like returns, billing disputes, and account issues. It has access to your backend systemsthrough custom Model Context Protocol (MCP) tools ( get_customer , lookup_order , process_refund ,escalate_to_human ). Your target is 80%+ first-contact resolution while knowing when to escalate.During testing, you find that when a customer says “I need a refund for my recent purchase,” the agent callsprocess_refund immediately—but populates the required order_id parameter with a plausible-looking butfabricated value instead of first calling lookup_order to retrieve the actual order ID. The refund call failsbecause the fabricated ID doesn’t exist.Which change directly addresses the root cause of the agent fabricating the order_id value?
Answer: A
-
You are building a customer support resolution agent using the Claude Agent SDK. The agent handles highambiguity requests like returns, billing disputes, and account issues. It has access to your backend systemsthrough custom Model Context Protocol (MCP) tools ( get_customer , lookup_order , process_refund ,escalate_to_human ). Your target is 80%+ first-contact resolution while knowing when to escalate.During testing, you find that when a customer says “I need a refund for my recent purchase,” the agent callsprocess_refund immediately—but populates the required order_id parameter with a plausible-looking butfabricated value instead of first calling lookup_order to retrieve the actual order ID. The refund call failsbecause the fabricated ID doesn’t exist.Which change directly addresses the root cause of the agent fabricating the order_id value?
-
Question 4
-
You are building a structured data extraction system using Claude. The system extracts information fromunstructured documents, validates the output using JavaScript Object Notation (JSON) schemas, andmaintains high accuracy. It must handle edge cases gracefully and integrate with downstream systems.The system needs to extract candidate information (name, contact details, skills, work experience, education)from uploaded resumes. The extracted data must strictly conform to a predefined JSON schema, as missingrequired fields or incorrect data types will cause downstream validation failures.What is the most reliable approach to ensure Claude’s output consistently matches the schema?
Answer: D
-
You are building a structured data extraction system using Claude. The system extracts information fromunstructured documents, validates the output using JavaScript Object Notation (JSON) schemas, andmaintains high accuracy. It must handle edge cases gracefully and integrate with downstream systems.The system needs to extract candidate information (name, contact details, skills, work experience, education)from uploaded resumes. The extracted data must strictly conform to a predefined JSON schema, as missingrequired fields or incorrect data types will cause downstream validation failures.What is the most reliable approach to ensure Claude’s output consistently matches the schema?
-
Question 5
-
You built an LLM-powered code-review tool that analyzes pull requests and returns structured findings. Each
finding is a JSON object containing file_path, line_number, issue_category—such as security or style—and
description. Developers can dismiss findings they consider unhelpful, and currently 35% of findings are
dismissed. You want to analyze these dismissals to understand what the system is getting wrong and improve
the prompts accordingly. What change to the output structure would best support this analysis?
Answer: B
-
You built an LLM-powered code-review tool that analyzes pull requests and returns structured findings. Each
finding is a JSON object containing file_path, line_number, issue_category—such as security or style—and
description. Developers can dismiss findings they consider unhelpful, and currently 35% of findings are
dismissed. You want to analyze these dismissals to understand what the system is getting wrong and improve
the prompts accordingly. What change to the output structure would best support this analysis?