PracticeVideosBlogFlashcards
All Practice Exams

Free Practice Questions for Claude Certified Architect - Foundations

Exam-style questions and explanations by OpenExamPrep.

✓ No registration✓ No credit card
100+ Questions
100% Free

Loading practice questions...

Same family resources

Explore More Anthropic Claude Certifications

Continue into nearby exams from the same family. Each card keeps practice questions, study guides, flashcards, videos, and articles in one place.

Claude Certified Associate - Foundations (CCAO-F)

Practice Questions114 questionsStudy Guide30 topicsCheat Sheet84 factsFlashcards50 cardsGames30 termsFeedVideos4 videosBlogs4 blogs

Claude Certified Developer - Foundations (CCDV-F)

Practice Questions164 questionsStudy Guide40 topicsCheat Sheet225 factsFlashcards50 cardsGames44 termsFeedVideos1 videoBlogs1 blog

Claude Certified Architect - Professional (CCAR-P)

Practice Questions100 questionsFeedVideos1 videoBlogs1 blog

More From This Family

Videos and articles for deeper review.

VideoCCA-F Practice Questions 2026: Current CCAR-F Blueprint, Scenarios, and Answer StrategyThis examVideoCCAR-F Exam Guide 2026: Claude Certified Architect – FoundationsThis examVideoCCAR-F vs CCA-F: Registration and Pearson VUE Guide for Claude Architect Foundations (2026)This examVideoWhich Claude Certification Should You Take in 2026? CCAO-F vs CCDV-F vs CCAR-F vs CCAR-PThis examArticleCCAR-F Exam Guide 2026: Claude Certified Architect – Foundations12 min readArticleWhich Claude Certification Should You Take in 2026? CCAO-F vs CCDV-F vs CCAR-F vs CCAR-P13 min readArticleCCA-F Practice Questions 2026: Current CCAR-F Blueprint, Scenarios, and Answer Strategy15 min readArticleCCAR-F vs CCA-F: Registration and Pearson VUE Guide for Claude Architect Foundations (2026)19 min read
Exam Review

Key Facts: Claude Certified Architect - Foundations Exam

720/1000

Passing Score

Anthropic exam guide v1.0

60

Exam Questions

Anthropic exam guide v1.0

120 min

Exam Time

Anthropic exam guide v1.0

5

Exam Domains

Anthropic exam guide v1.0

$125

Exam Fee

Anthropic certification page

Mar 2026

Launch

Anthropic Claude Partner Network announcement

CCAR-F requires a scaled score of 720 out of 1000 across 60 multiple-choice and multiple-response questions in 120 minutes. The exam costs $125, is valid for 12 months, and is delivered online-proctored or at a Pearson VUE test center.

Sample Claude Certified Architect - Foundations Practice Questions

Try these sample questions to review concepts for the Claude Certified Architect - Foundations exam. Each question includes a detailed explanation. Start the interactive quiz above for the full 100+ question experience with AI tutoring.

1A team is building a customer-support agent. They want the simplest reliable architecture and only plan to add complexity if needed. According to Anthropic's guidance on building effective agents, what should they start with?
A.A multi-agent swarm where dozens of agents negotiate the answer
B.The simplest solution possible, such as a single LLM call with retrieval and in-context examples, adding agentic steps only when simpler approaches fall short
C.An orchestrator-worker system with at least five specialized workers from day one
D.A fine-tuned model dedicated to support before writing any prompts
Explanation: Anthropic's 'Building Effective Agents' guidance is explicit: find the simplest solution possible and only increase complexity when needed. Many use cases are solved by a single optimized LLM call with retrieval and examples, and full agentic systems should be reserved for when simpler patterns demonstrably fall short. This keeps systems transparent, cheaper, and easier to evaluate.
2An architect needs to classify incoming requests and send each to a specialized prompt: billing questions to one prompt, technical issues to another, and general questions to a third. Which workflow pattern best fits this requirement?
A.Parallelization
B.Evaluator-optimizer
C.Prompt chaining
D.Routing
Explanation: Routing classifies an input and directs it to a specialized downstream prompt or model. It works well when there are distinct categories better handled separately and when classification can be done accurately. Separating concerns this way lets you optimize each path independently without one prompt trying to handle everything.
3A coding task requires generating a draft, then critiquing it against quality criteria, then refining it, repeating until the critique passes. Which pattern describes this, and what is the key design principle?
A.Orchestrator-worker, with one orchestrator spawning all critics at once
B.Routing, where a classifier picks the best draft from several
C.Evaluator-optimizer, separating generation and evaluation into distinct LLM calls with an adversarial evaluator stance
D.Parallelization, generating many drafts and averaging them
Explanation: The evaluator-optimizer pattern uses one call to generate and another to evaluate and give feedback, looping until criteria are met. Anthropic recommends separating generation from evaluation into distinct calls and giving the evaluator an adversarial stance so it finds real problems. This is one of the highest-leverage improvements for quality-sensitive tasks.
4A research agent must handle tasks where the number and nature of subtasks cannot be predicted in advance: a central model must dynamically break the work down, delegate pieces, and synthesize the results. Which pattern is the best fit?
A.Prompt chaining with a fixed sequence of steps
B.Orchestrator-worker
C.Sectioning a single prompt into headers
D.Routing to one of three static categories
Explanation: In the orchestrator-worker pattern, a central LLM dynamically decomposes a task, delegates subtasks to worker LLMs, and synthesizes their results. It is specifically suited to problems where you cannot predict the subtasks ahead of time, unlike fixed chains or static routing. The orchestrator adapts the plan based on what it discovers.
5An architect wants to distinguish a 'workflow' from an 'agent' in Anthropic's terminology. Which statement is correct?
A.Workflows let the LLM dynamically direct its own process and tool usage; agents follow predefined code paths
B.Workflows and agents are identical terms with no practical difference
C.Agents never use tools, while workflows always do
D.Workflows orchestrate LLMs and tools through predefined code paths; agents dynamically direct their own processes and tool usage
Explanation: Anthropic defines workflows as systems where LLMs and tools are orchestrated through predefined code paths, while agents are systems where the LLM dynamically directs its own processes and tool usage, controlling how it accomplishes the task. The distinction guides architecture choices: workflows offer predictability, agents offer flexibility for open-ended problems.
6A latency-sensitive content pipeline must run a translation, a sentiment analysis, and a toxicity check on the same input, all independent of each other, then combine the results. Which pattern minimizes end-to-end latency?
A.Prompt chaining the three checks sequentially
B.Parallelization (sectioning), running the three independent calls concurrently and aggregating
C.Evaluator-optimizer looping over each check
D.A single giant prompt asked to do all three plus self-grade
Explanation: Parallelization runs independent subtasks simultaneously and aggregates their outputs. When subtasks do not depend on one another, sectioning them into concurrent calls reduces wall-clock latency compared to sequential chaining. It can also improve focus, since each call handles one concern.
7An agent autonomously calls tools in a loop. Anthropic recommends giving the agent guardrails and a way to stop. Which design element most directly bounds runaway loops in a production agent?
A.Setting a maximum number of turns/iterations and clear stopping conditions
B.Increasing max_tokens to the model's maximum on every call
C.Disabling all tool use after the first call
D.Removing the system prompt to reduce token cost
Explanation: Autonomous agents run in a loop, so you must bound them with stopping conditions and a maximum number of turns to prevent runaway cost or infinite loops. The Agent SDK exposes options like maxTurns for exactly this. Clear stop criteria are a core reliability guardrail for agentic systems.
8When designing a multi-step agent, Anthropic emphasizes 'showing the agent's planning steps' and keeping the design simple. What is the primary architectural benefit of this transparency?
A.It guarantees the agent never makes a mistake
B.It eliminates the need for any evaluation or testing
C.It makes the agent's behavior easier to debug, audit, and trust, and easier to course-correct
D.It allows the model to skip tool calls entirely
Explanation: Anthropic recommends prioritizing transparency by explicitly showing the agent's planning steps. Exposing the plan makes behavior observable, so engineers can debug, audit, and intervene when the agent goes off track. Transparency is a core operability principle for production agents, not a correctness guarantee.
9A team uses the Claude Agent SDK and spawns several subagents in parallel for a large refactor. Each subagent does deep analysis. What is the main architectural advantage of using separate subagents here?
A.Subagents share one context window, so all analysis is visible to every agent
B.Subagents automatically inherit all parent permissions, removing any approval prompts
C.Subagents cannot use tools, which makes them faster
D.Each subagent gets its own context window, so one subagent's exploration does not pollute the others' or the main context
Explanation: In the Claude Agent SDK, each subagent runs with its own isolated context window, so verbose exploration in one subagent does not consume or pollute the main conversation or sibling subagents. This context preservation is a primary reason to use subagents for parallelizable, context-heavy subtasks. The main agent can then synthesize concise results.
10An architect must choose a model for a high-volume, latency-sensitive classification step inside a larger agent, where most decisions are simple. Which model selection strategy aligns with Anthropic's guidance?
A.Always use the largest Opus model everywhere for maximum quality
B.Use a fine-tuned model because base models cannot classify
C.Use a faster, cheaper model such as Haiku for the simple high-volume step, reserving larger models for genuinely hard reasoning steps
D.Route every step to a random model to balance load
Explanation: Anthropic's model lineup is tiered: Haiku for fast, low-cost, high-volume tasks; Sonnet for balanced general work; Opus for the hardest reasoning. A good architecture matches each step to the cheapest model that meets the quality bar, using Haiku for simple high-volume classification and reserving Opus for complex reasoning. This optimizes both cost and latency.

About the Claude Certified Architect - Foundations Exam

Claude Certified Architect - Foundations (current Pearson code CCAR-F; formerly CCA-F) is Anthropic's architecture certification for designing, deploying, and maintaining production-grade Claude applications. The exam covers agentic architecture and orchestration (27%), tool design and MCP integration (18%), Claude Code configuration and workflows (20%), prompt engineering and structured output (20%), and context management and reliability (15%).

Exam sponsor: Anthropic. The requirements and fees below concern the certification or admission exam, separate from our free practice resources.

Questions

60 questions

Time Limit

120 minutes

Passing Score

720/1000

Exam / Certification Fees

$125

Exam sponsor website

Fees, eligibility, and exam policies can change. Confirm them with the exam sponsor before applying or paying.

Our practice resources: topics covered

We aim to reflect publicly available exam outlines and topic information in our study resources. Coverage, format, and difficulty may differ from the actual exam, and we cannot guarantee that every detail is accurate or current. Confirm exam requirements, fees, and policies with the official exam sponsor.

27%

Agentic Architecture & Orchestration

Design agentic loops that continue on stop_reason tool_use and end on end_turn, orchestrate coordinator-subagent systems with explicit context passing and parallel subagents, enforce workflow order with programmatic prerequisites and structured handoffs, apply Agent SDK hooks to intercept tool calls and normalize results, choose fixed prompt chains or adaptive decomposition, and manage session resumption and forking

18%

Tool Design & MCP Integration

Write tool descriptions with clear boundaries, return structured MCP errors with an error category and retryable flag, scope tools to each agent's role and configure tool_choice, configure MCP servers in project .mcp.json or user ~/.claude.json with environment variable expansion, expose content catalogs as resources, and pick the right built-in tool

20%

Claude Code Configuration & Workflows

Configure the CLAUDE.md hierarchy with @import and .claude/rules/, apply path-specific rules with glob patterns, build custom slash commands and skills (context: fork, allowed-tools, argument-hint), choose plan mode or direct execution, refine iteratively with examples, tests and the interview pattern, and run Claude Code in CI/CD with -p, --output-format json and --json-schema

20%

Prompt Engineering & Structured Output

Write explicit criteria that reduce false positives, add targeted few-shot examples, enforce structured output with tool use and JSON schemas (tool_choice, nullable and enum fields), build validation-retry loops with error feedback, decide when the Message Batches API fits, and design multi-instance and multi-pass reviews

15%

Context Management & Reliability

Preserve critical facts across long interactions, set escalation triggers and resolve ambiguity, propagate structured errors across multi-agent systems, manage context while exploring large codebases, design human review with confidence calibration and stratified sampling, and keep claim-source provenance in multi-source synthesis

Preparing for the Claude Certified Architect - Foundations Exam

What You Need to Know

  • Passing score: 720/1000
  • Exam length: 60 questions
  • Time limit: 120 minutes
  • Exam / certification fees: $125 Official sources

Using Our Practice Resources

  • Work through all 100 available questions
  • Review every answer and explanation
  • Track weak areas and revisit them
  • Use our AI tutor for tough concepts

Claude Certified Architect - Foundations: Suggested Study Strategy

1Prioritize the highest-weighted domain (27%): agentic architecture and orchestration with the Claude Agent SDK, subagents, and multi-agent patterns
2Know when a fixed prompt chain beats adaptive decomposition, and when a hook or programmatic prerequisite must enforce an order that a prompt can only suggest
3Learn the agentic loop exactly: continue on stop_reason tool_use, stop on end_turn, and treat iteration caps or parsing the reply text as anti-patterns for the main stopping rule
4Know MCP configuration cold: project-scoped .mcp.json for shared servers, user-scoped ~/.claude.json for personal ones, environment variable expansion for secrets, structured isError responses, and resources for content catalogs
5Practice structured output: tool use with JSON schemas, tool_choice auto vs any vs forced, nullable fields that prevent fabricated values, and validation-retry loops for semantic errors a schema cannot catch
6Study Claude Code configuration: the CLAUDE.md hierarchy, @import and .claude/rules/ path scoping, skill frontmatter (context: fork, allowed-tools, argument-hint), plan mode vs direct execution, and -p with --output-format json and --json-schema in CI
7Know the reliability patterns: case-facts blocks and trimmed tool outputs for long sessions, explicit escalation triggers instead of sentiment, error propagation with partial results, confidence calibration with stratified sampling, and claim-source provenance
8Complete 100+ practice questions across all five domains and review every wrong answer until you can explain the architecture principle behind the correct choice

Frequently Asked Questions

What is the CCAR-F passing score?

CCAR-F requires a scaled score of 720 on a 100-1000 range. The current exam has 60 multiple-choice and multiple-response questions in 120 minutes.

How much does the CCAR-F exam cost?

The current CCAR-F exam fee is $125 USD before any partner-tier discount. Each retake is charged at the full exam fee. The credential is valid for 12 months; renewing on time, before it lapses, is a free, non-proctored assessment on the Anthropic Partner Academy, while a lapsed credential means retaking the full exam at the full fee.

What does the CCAR-F exam cover?

CCAR-F covers agentic architecture (27%), tool design and MCP (18%), Claude Code (20%), prompt engineering and structured output (20%), and context management and reliability (15%). Questions are set in 4 scenarios drawn at random from a bank of 6.

What prior experience do I need for CCAR-F?

Anthropic lists no formal prerequisite, but certification is open only to people at Claude Partner Network organizations, who register with a partner company email address. The exam guide targets solution architects with roughly six or more months of practical Claude experience.

How long should I study for CCAR-F?

Anthropic does not publish a study-time figure; the exam guide lists recommended experience only. Spend significant time building: configure MCP servers, design multi-agent workflows with the Claude Agent SDK, and practice structured-output and reliability patterns. The Anthropic Partner Academy lists free prep courses for this exam, including Building with the Claude API, Claude Code in Action, and Introduction to Model Context Protocol.

What is MCP and why does it matter for CCAR-F?

Model Context Protocol connects Claude to external tools, resources, and prompts. Tool Design and MCP Integration is 18% of the official blueprint.

How is CCAR-F delivered and when can I retake it?

The exam is online-proctored or delivered at a Pearson VUE test center. Failed attempts require 14-, 30-, then 90-day waits, with four attempts per rolling 12 months.