Prepare for the Claude Certified Architect Foundations certification with comprehensive CCAR-F practice tests and mock exams. Test your knowledge with realistic, scenario-based questions, review detailed explanations, and strengthen your preparation across the key architecture and exam domains. Start practicing today and build your confidence for the CCAR-F exam.

CCAR-F Practice Exam Questions

Questions: 152

Updated: Sep 25, 2026

Package *

PDF

$69.00

Special Price

152 Q&A with explanations
Printable PDF, any device
Offline access, no login

Select Package Selected

Pass guarantee included

Exam Engin

$79.00

Special Price

Timed mock exams
Domain score analytics
Web-based + downloadable

Select Package Selected

Pass guarantee included

PDF + Exam Engin

$89.00

Special Price

152 Q&A explained
PDF study format
Testing engine
Timed & practice modes
Domain score analytics
Web + downloadable

Select Package Selected

Pass guarantee included

Duration *

1 month

3 month $15.00

6 month $30.00

$0.00

More Information

Anthropic	Anthropic
Exam name	Claude Certified Architect – Foundations
Exam Code	CCAR-F
Number of Question	152
Update at	Sep 25, 2026

Description

2026 Complete Guide & Mock System

Claude Certified Architect – Foundations (CCAR-F)

Everything you need to master the CCAR-F exam — structure, blueprint domains, production scenarios, fees, and tactical study plans, paired with scenario-grounded Claudemock™ practice simulations.

01. What is the CCAR-F Certification?

The Claude Certified Architect – Foundations (CCAR-F) is Anthropic's foundational technical credential for professionals who design, build, and deploy production-grade systems on Claude. Built for solutions architects, software engineers, AI engineers, and technical leads, it assesses your ability to make sound architectural decisions across agentic orchestration, tool and MCP design, Claude Code workflows, structured output, and context management.

Originally launched in March 2026 under the community shorthand "CCA-F", the exam moved to Pearson VUE on 30 June 2026 (online proctoring through OnVUE or physical testing centers). Exam content was unchanged by the move. Passing candidates earn a verified digital Credly badge valid for 12 months, renewable through a free, non-proctored annual assessment.

Business / Non-Dev

CCAO-F

Foundations for day-to-day business productivity, prompting, & data governance.

Developer

CCDV-F

API integration, Agent SDK implementation, & programmatic tool-calling.

Architecture

CCAR-F

Foundational agentic architecture, MCP tool design, Claude Code, & reliability.

Architecture

CCAR-P

Advanced multi-tenant systems, high-scale agent orchestration, & governance.

02. Exam Format & Specifications

Questions	60 scenario-based items (Multiple-choice & multiple-response)
Duration	120 minutes (~2 minutes per item)
Passing Benchmark	Scaled score of 720 (range: 100–1,000)
Scenario Structure	4 of 6 production scenarios randomly selected per sitting; all items anchored to those scenarios
Testing Delivery	Pearson VUE OnVUE (remote proctored) or Authorized Test Centers
Registration Fee	$125 USD per attempt (No complimentary retakes; eligible partner-tier discounts shown at checkout)
Prerequisites	Corporate association with a Claude Partner Network tier (Free "Registered" tier eligible). ~6 months hands-on with the Claude API, Agent SDK, Claude Code & MCP recommended
Cooldown / Retake Policy	Attempt 2: 14-day wait \| Attempt 3: 30 days \| Attempt 4: 90 days (Max 4/year)
Validity & Renewal	12 months; free, non-proctored annual renewal module

03. Exam Domains & Blueprint Weights

Nearly half (47%) of the total exam weight centers on Agentic Architecture and Claude Code Configuration. Memorizing API parameters or definitions will not yield a passing score — you must make production-grade design decisions inside realistic deployment scenarios.

Domain	Weight	Core Evaluated Skills
1. Agentic Architecture & Orchestration	27%	Agent loops, task decomposition, hub-and-spoke subagent coordination, human-in-the-loop escalation.
2. Tool Design & MCP Integration	18%	Tool descriptions & schemas, tool selection errors, MCP primitives (tools, resources, prompts), error responses.
3. Claude Code Configuration & Workflows	20%	CLAUDE.md hierarchy, custom slash commands, Skills frontmatter, hooks, CI/CD & headless integration.
4. Prompt Engineering & Structured Output	20%	Explicit criteria, few-shot conditioning, JSON schema via tool use, validation-retry loops, batch & multi-pass review.
5. Context Management & Reliability	15%	Context window budgeting, prompt caching, conversation compaction, graceful degradation, long-session reliability.

The 6 Production Scenarios (4 appear per sitting)

Customer Support Resolution Agent — intent routing, knowledge retrieval, escalation on low confidence.

Code Generation with Claude Code — codebase reading, proposed changes, command execution.

Multi-Agent Research System — coordinator and subagents, context delegation, synthesis.

Developer Productivity Tooling — shared conventions, custom commands, team workflows.

Claude Code for CI/CD — non-interactive runs, automated review, pipeline guardrails.

Structured Data Extraction — schema enforcement, validation, batch processing.

04. Ideal Candidate Profile

Built For

Solutions architects designing Claude-powered systems.
Software & AI engineers shipping agents to production.
Technical leads standardizing Claude Code across teams.
Consultants delivering MCP and Agent SDK integrations.

Out of Scope (Choose CCAO-F or CCAR-P)

Business users relying on Claude for daily productivity (CCAO-F).
Non-technical managers without API or Claude Code experience.
Architects of high-scale, multi-tenant enterprise platforms (CCAR-P).

05. Why Candidates Fail & How Claudemock Solves It

Scenario Context Switching

Four production scenarios in 120 minutes force rapid shifts between architectures.

Claudemock: Randomized 4-of-6 scenario mocks with full 120-min timers.

Prompt vs. Programmatic Traps

Many distractors "fix" a reliability problem with prompt wording instead of enforced controls.

Claudemock: Full rationale breakdowns explain exactly why traps fail.

Claude Code Specifics

CLAUDE.md scopes, Skills frontmatter, and CI flags are tested in precise detail.

Claudemock: Domain 3 drills internalize configuration details without hints.

06. Recommended 4-Week Study Sprint

Week 1

Core Architecture & Foundations

Read the official CCAR-F exam guide and its sample questions. Complete Anthropic Academy courses on the Claude API, MCP (intro and advanced), Claude Code, subagents, and Agent Skills.

Week 2

Domain-by-Domain Drilldowns

Complete all practice modules on Claudemock.com, prioritizing Agentic Architecture and Claude Code Configuration. Read all feedback rationales, especially for correct answers.

Week 3

Hands-on Sandbox Build

Build a tool-using agent loop with an escalation path, write a small MCP server, configure project-level CLAUDE.md, custom commands and hooks, and run Claude Code non-interactively in a CI job.

Week 4

Simulated Full Mocks & Pacing

Complete two timed, closed-book Claudemock exams in single sittings. Verify your domain radar chart stays consistently above 85% before scheduling on Pearson VUE.

07. Sample Examination Questions

Original practice questions matching official CCAR-F depth and difficulty.

Domain 1: Agentic Architecture & Orchestration

A customer support agent can issue refunds through a tool. Policy requires human approval for any refund above $500. In testing, the agent occasionally issues large refunds despite a clear system prompt rule. What is the most reliable fix?

A. Repeat the refund rule in capital letters at the end of the system prompt

B. Enforce the threshold in code so refunds above $500 are blocked and routed to a human approval step

C. Add few-shot examples of the agent declining large refunds

D. Switch to a larger model with stronger instruction following

Correct Answer: B.

Prompt instructions are guidance, not guarantees. When a business rule has financial or compliance consequences, it must be enforced programmatically at the tool boundary, with a deterministic escalation path to a human.

Domain 2: Tool Design & MCP Integration

An agent has two tools, search_orders and get_order, and frequently calls the wrong one. Both have one-line descriptions. Which two changes most directly improve tool selection? (Select 2)

[A] Expand each description to state its purpose, expected inputs, and when to use it instead of the other tool

[B] Increase the sampling temperature so the agent explores both tools

[C] Make input schemas clearly distinct (e.g. exact order ID vs. customer filters)

[D] Add a third general-purpose orders tool as a fallback

Correct Answers: A and C.

Claude selects tools from their descriptions and schemas. Clear, differentiated descriptions and distinct input shapes remove ambiguity. Adding randomness or overlapping tools makes misselection worse.

Domain 3: Claude Code Configuration & Workflows

A tech lead wants every engineer using Claude Code in a repository to follow the same testing and naming conventions automatically, with changes reviewed like any other code. Where should these conventions live?

A. Each engineer's user-level ~/.claude/CLAUDE.md

B. A project-level CLAUDE.md committed to the repository

C. A pinned message in the team chat channel

D. Pasted at the start of each Claude Code session

Correct Answer: B.

A project-level CLAUDE.md is loaded automatically for everyone working in the repo and is version-controlled, so conventions are shared and reviewable. User-level files apply only to one person; manual pasting is inconsistent.

08. Frequently Asked Questions

It is not a definitions quiz. Every item places you as the architect of a specific production system and asks for the best design decision. Candidates without hands-on experience building with the Claude API, Claude Code, and MCP typically find it demanding under the 120-minute limit.

Yes. Registration requires access through Anthropic Partner Academy. Candidates must link an organization enrolled in the Claude Partner Network (the free "Registered" tier is sufficient) before Pearson VUE checkout unlocks.

Yes. CCA-F was the common label when the exam launched in March 2026. CCAR-F is the official code used since the move to Pearson VUE. The content is the same, but older guides may still list the earlier $99 fee, ProctorFree delivery, or the retired official practice exam — ignore those details.

No. Anthropic's Certification Exam Policy explicitly prohibits AI assistance, and Pearson VUE exams are closed-book and monitored via OnVUE webcam supervision or on-site proctors.

You must wait 14 days before taking Attempt 2, 30 days before Attempt 3, and 90 days before Attempt 4 (limited to 4 attempts within any rolling 12-month period). Retakes require payment of the full $125 fee.

Ready to Pass CCAR-F on Your First Attempt?

Sharpen your timing, work through realistic production scenarios, and track domain-level readiness with Claudemock™ practice exams.

Start Mock Exam Download Study Guide

Reviews

Write Your Own Review

Only registered users can write reviews. Please Sign in or create an account

Practice Preview

Loading questions…

No practice questions are available for this exam yet.