BrainyBeeBrainyBee
ExploreBlogStart Studying
Home›Explore›Claude Certified Architect - Foundations (CCAR-F)
🤖

🤖 Anthropic

Free Claude Certified Architect - Foundations (CCAR-F) Study Resources

Get certified on Claude — agentic loops, MCP tool design, Claude Code and structured output, weighted exactly as the exam is. 253 questions set inside the six real exam scenarios, four full-length papers that reproduce its four-of-six draw, 240 flashcards and 12 study notes. All grounded in Anthropic's own docs. Free.

253
Practice Questions
4
Mock Exams
12
Study Notes
240
Flashcard Decks
482
Source Materials
Start Studying — Free

On This Page

  • Study Notes (12)
  • Practice Questions (15)
  • Flashcards (30)
  • Related Study Resources

Claude Certified Architect - Foundations (CCAR-F) Study Notes & Guides

12 AI-generated study notes covering the full Claude Certified Architect - Foundations (CCAR-F) curriculum. Showing 10 complete guides below.

Exam map194 words

CCAR-F: how the exam is dealt

Claude Certified Architect - Foundations (CCAR-F)

Read full article

CCAR-F: how the exam is dealt

The Claude Certified Architect – Foundations exam asks one kind of question over and over: given this production system, which design holds up? It is not a recall test about API parameters. Almost every item puts a working system in front of you, breaks something about it, and offers four plausible repairs.

The exam at a glance

Items
60
Time limit
120 minutes
Structure
4 scenarios drawn from a bank of 6
Item format
Multiple-choice and multiple-response; each item states how many to select
Passing score
720 scaled, on 100–1,000
Delivery
Proctored — online or test centre, via Pearson VUE
Validity
12 months from the date awarded

The blueprint

Five domains, weighted. The weights are approximate proportions of scored items, and they are the single best guide to where your study hours belong.

Domain weights

1 · Agentic Architecture & Orchestration
27%
2 · Tool Design & MCP Integration
18%
3 · Claude Code Configuration & Workflows
20%
4 · Prompt Engineering & Structured Output
20%
5 · Context Management & Reliability
15%

Domain 1 is the largest single block and Domain 5 the smallest, but the spread is narrow — there is no domain you can afford to skip, and no domain that will carry you on its own.

What the scenario draw means for you

Four of six, and you do not choose which four

Every item you see sits inside one of six published production contexts, and four of them are dealt at random on the day. Two scenarios you prepared will not appear — and you will not know which two until you are sitting the exam. Preparing five of six is preparing to fail on the sixth. The scenario briefs in this hive cover all six for exactly this reason.

The domains do not distribute evenly across the scenarios either. Prompt engineering and structured output live mostly in the extraction and CI contexts; the multi-agent and support contexts carry orchestration and context management. So the draw moves your domain mix, which is why the practice papers here draw four scenarios apiece rather than dealing a fixed blend.

A study sequence that works

From cold to exam-ready

  1. 1

    Read the six scenario briefs first

    They are short, and every question you will ever see is framed by one of them. Meeting a scenario for the first time in a practice paper wastes the reading time on orientation instead of on the question.

What the exam will not reward

The plausible answer is usually the prompt-based one

A recurring pattern across all five domains: the wrong answer improves the prompt, and the right answer changes the structure. Hooks over instructions, prerequisite gates over stated ordering, scoped tool sets over told-not-to, schemas over asking nicely. When one option makes a failure impossible and another makes it less likely, the exam wants the first.

Where the facts on this page come from

Every figure above is quoted from the official CCAR-F exam guide v1.0 (effective July 2026) — sections 3, 4 and 5. Registration, retake and accommodation policies change independently of this guide; check the Anthropic Partner Academy page before you book.

Scenario brief225 words

Scenario 1: Customer Support Resolution Agent

Claude Certified Architect - Foundations (CCAR-F)

Read full article

Scenario 1: Customer Support Resolution Agent

You are building a customer support resolution agent using the Claude Agent SDK. The agent handles high-ambiguity requests like returns, billing disputes, and account issues. It has access to your backend systems through 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.

Passage quoted verbatim from the official CCAR-F exam guide v1.0, section 5.

This scenario on the exam

Scenario
1 of 6 — four are dealt at random
Primary domains
3
Questions in this hive
45

Primary domains

  • Agentic Architecture & Orchestration
  • Tool Design & MCP Integration
  • Context Management & Reliability

What the exam asks inside this context

  • Enforcement before money moves. Identity verification ahead of a refund is the canonical prerequisite-gate item, and the correct answer is programmatic every time. A refund above a threshold gets intercepted and rerouted to escalation, not talked out of.

  • Tool descriptions with boundaries. The four named tools are deliberately close together, and questions turn on what a description fails to say — a threshold, an input format, an edge case.

  • When to escalate. Three legitimate triggers: the customer asks for a human, policy is silent or has an exception, or the agent cannot make progress. Frustration is not one of them, and neither is a low self-reported confidence.

  • Facts that survive a long conversation. Amounts, dates, order numbers and statuses belong in a persistent case-facts block outside the summarized history, because summarization is what eats them.

Where this scenario catches people

The 80% first-contact resolution target reads like a reason to keep trying. It is not — an explicit request for a human is honoured immediately, and a policy gap escalates rather than being resolved by inference from an adjacent rule. The target is context, not an instruction to resist escalation.

Read the passage once, properly

Every item in this block shares this passage, so the two minutes you spend on it are amortised across the whole block — and the details in it are load-bearing. The named tools, the stated target, the tools the agent is said to have: each of those has been the deciding fact in a question.

What is quoted, and what is argued

The passage and the primary-domain list are the exam guide's own words. Everything after them is this hive's reading of what the domains imply for this context — useful preparation, not an official statement about what the exam contains. The guide publishes 60 items and 4 scenarios per sitting and does not publish a per-scenario count, so none is quoted here.

Scenario brief185 words

Scenario 2: Code Generation with Claude Code

Claude Certified Architect - Foundations (CCAR-F)

Read full article

Scenario 2: Code Generation with Claude Code

You are using Claude Code to accelerate software development. Your team uses it for code generation, refactoring, debugging, and documentation. You need to integrate it into your development workflow with custom slash commands, CLAUDE.md configurations, and understand when to use plan mode vs direct execution.

Passage quoted verbatim from the official CCAR-F exam guide v1.0, section 5.

This scenario on the exam

Scenario
2 of 6 — four are dealt at random
Primary domains
2
Questions in this hive
50

Primary domains

  • Claude Code Configuration & Workflows
  • Context Management & Reliability

What the exam asks inside this context

  • Configuration scope. Which CLAUDE.md level reaches a teammate, which slash command directory is shared, and why a new joiner is missing instructions that work perfectly for everyone else.

  • Plan mode against direct execution. Scope decides it. A single-file fix with a clear stack trace is executed; a migration across forty-five files is planned first.

  • Skills and their frontmatter. context: fork for verbose or exploratory work, allowed-tools to keep a skill from doing damage, argument-hint for a missing parameter.

  • Iterative refinement. Concrete input/output examples and failing tests, rather than more description. Interacting problems go in one message; independent ones are better fixed one at a time.

Where this scenario catches people

This scenario invites answers that add prose to CLAUDE.md. Ask instead whether the guidance should load at all times — if it should not, it belongs in a skill or in a path-scoped rule, and a growing always-loaded file is the defect rather than the fix.

Read the passage once, properly

Every item in this block shares this passage, so the two minutes you spend on it are amortised across the whole block — and the details in it are load-bearing. The named tools, the stated target, the tools the agent is said to have: each of those has been the deciding fact in a question.

What is quoted, and what is argued

The passage and the primary-domain list are the exam guide's own words. Everything after them is this hive's reading of what the domains imply for this context — useful preparation, not an official statement about what the exam contains. The guide publishes 60 items and 4 scenarios per sitting and does not publish a per-scenario count, so none is quoted here.

Scenario brief199 words

Scenario 3: Multi-Agent Research System

Claude Certified Architect - Foundations (CCAR-F)

Read full article

Scenario 3: Multi-Agent Research System

You are building a multi-agent research system using the Claude Agent SDK. A coordinator agent delegates to specialized subagents: one searches the web, one analyzes documents, one synthesizes findings, and one generates reports. The system researches topics and produces comprehensive, cited reports.

Passage quoted verbatim from the official CCAR-F exam guide v1.0, section 5.

This scenario on the exam

Scenario
3 of 6 — four are dealt at random
Primary domains
3
Questions in this hive
47

Primary domains

  • Agentic Architecture & Orchestration
  • Tool Design & MCP Integration
  • Context Management & Reliability

What the exam asks inside this context

  • What a subagent inherits. Nothing. Not the parent conversation, not earlier findings, not memory between invocations — so the synthesis subagent gets the complete prior findings in its prompt or it does not have them.

  • Decomposition that covers. Slice a topic too narrowly and every slice is complete while the cross-cutting question is owned by nobody. Partition scope, by subtopic or by source type, before the subagents start.

  • Errors that a coordinator can act on. Failure type, what was attempted, partial results, alternatives. A generic 'search unavailable' is technically true and useless.

  • Provenance through synthesis. Claim-source mappings with URLs, document names and excerpts, preserved and merged rather than compressed away — plus dates, so a year-old figure is not read as a contradiction.

Where this scenario catches people

Conflicting statistics from two credible sources are not a problem to resolve. The report annotates both with attribution and lets the reader see the disagreement; picking one value presents a contested number as settled, which is the specific failure this scenario is built to catch.

Read the passage once, properly

Every item in this block shares this passage, so the two minutes you spend on it are amortised across the whole block — and the details in it are load-bearing. The named tools, the stated target, the tools the agent is said to have: each of those has been the deciding fact in a question.

What is quoted, and what is argued

The passage and the primary-domain list are the exam guide's own words. Everything after them is this hive's reading of what the domains imply for this context — useful preparation, not an official statement about what the exam contains. The guide publishes 60 items and 4 scenarios per sitting and does not publish a per-scenario count, so none is quoted here.

Scenario brief197 words

Scenario 4: Developer Productivity with Claude

Claude Certified Architect - Foundations (CCAR-F)

Read full article

Scenario 4: Developer Productivity with Claude

You are building developer productivity tools using the Claude Agent SDK. The agent helps engineers explore unfamiliar codebases, understand legacy systems, generate boilerplate code, and automate repetitive tasks. It uses the built-in tools (Read, Write, Bash, Grep, Glob) and integrates with Model Context Protocol (MCP) servers.

Passage quoted verbatim from the official CCAR-F exam guide v1.0, section 5.

This scenario on the exam

Scenario
4 of 6 — four are dealt at random
Primary domains
3
Questions in this hive
30

Primary domains

  • Tool Design & MCP Integration
  • Claude Code Configuration & Workflows
  • Agentic Architecture & Orchestration

What the exam asks inside this context

  • Which built-in for which job. Grep searches contents, Glob matches paths, Edit needs a unique anchor and Read plus Write is the fallback when there is not one.

  • Learning a codebase without reading it all. Grep to entry points, Read to follow imports and trace flows. Reading everything up front fills the context with files the task never touches.

  • Scoped tool sets. An agent holding explore, edit and deploy together will eventually edit during an investigation. The fix is the tool list, not an instruction.

  • MCP servers alongside the built-ins. Project versus user scope, environment variable expansion for tokens, and a description good enough that the agent does not fall back to Grep.

Where this scenario catches people

Tracing a function through wrapper modules is the item people lose here. Searching the original name finds the wrappers and misses every call that goes through them — identify all the exported names first, then search for each.

Read the passage once, properly

Every item in this block shares this passage, so the two minutes you spend on it are amortised across the whole block — and the details in it are load-bearing. The named tools, the stated target, the tools the agent is said to have: each of those has been the deciding fact in a question.

What is quoted, and what is argued

The passage and the primary-domain list are the exam guide's own words. Everything after them is this hive's reading of what the domains imply for this context — useful preparation, not an official statement about what the exam contains. The guide publishes 60 items and 4 scenarios per sitting and does not publish a per-scenario count, so none is quoted here.

Scenario brief168 words

Scenario 5: Claude Code for Continuous Integration

Claude Certified Architect - Foundations (CCAR-F)

Read full article

Scenario 5: Claude Code for Continuous Integration

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. You need to design prompts that provide actionable feedback and minimize false positives.

Passage quoted verbatim from the official CCAR-F exam guide v1.0, section 5.

This scenario on the exam

Scenario
5 of 6 — four are dealt at random
Primary domains
2
Questions in this hive
36

Primary domains

  • Claude Code Configuration & Workflows
  • Prompt Engineering & Structured Output

What the exam asks inside this context

  • Running non-interactively. -p, or the job hangs. Then --output-format json with --json-schema, because inline PR comments need parseable findings rather than prose to scrape.

  • Precision over emphasis. Explicit categorical criteria — report bugs and security, skip minor style — rather than asking for high-confidence findings only.

  • Independent review. The session that generated the code is the worst reviewer of it, because it holds the reasoning that produced the mistake.

  • Repeat runs. Prior findings go into context with an instruction to report only new or unaddressed issues, or every re-run posts the same comments again.

Where this scenario catches people

False positives are the subject of this scenario, and the tempting fix — a confidence filter — is the wrong one. What works is a categorical criterion, few-shot examples pairing acceptable patterns with genuine issues, and temporarily disabling a noisy category while its prompt is repaired.

Read the passage once, properly

Every item in this block shares this passage, so the two minutes you spend on it are amortised across the whole block — and the details in it are load-bearing. The named tools, the stated target, the tools the agent is said to have: each of those has been the deciding fact in a question.

What is quoted, and what is argued

The passage and the primary-domain list are the exam guide's own words. Everything after them is this hive's reading of what the domains imply for this context — useful preparation, not an official statement about what the exam contains. The guide publishes 60 items and 4 scenarios per sitting and does not publish a per-scenario count, so none is quoted here.

Scenario brief185 words

Scenario 6: Structured Data Extraction

Claude Certified Architect - Foundations (CCAR-F)

Read full article

Scenario 6: Structured Data Extraction

You are building a structured data extraction system using Claude. The system extracts information from unstructured documents, validates the output using JavaScript Object Notation (JSON) schemas, and maintains high accuracy. It must handle edge cases gracefully and integrate with downstream systems.

Passage quoted verbatim from the official CCAR-F exam guide v1.0, section 5.

This scenario on the exam

Scenario
6 of 6 — four are dealt at random
Primary domains
2
Questions in this hive
45

Primary domains

  • Prompt Engineering & Structured Output
  • Context Management & Reliability

What the exam asks inside this context

  • Guaranteed shape. tool_use with a JSON schema, and tool_choice set so a tool is actually called — 'any' when the document type is unknown, forced when one extraction must run first.

  • Schema design against fabrication. Optional and nullable fields where the source may be silent; an 'unclear' enum value and 'other' plus a detail string where the categories may not fit.

  • Validation that catches meaning. Schemas remove syntax errors and leave semantic ones. calculated_total beside stated_total, and a conflict_detected boolean, put the check in the output shape.

  • Retry, and knowing when not to. Retry with the specific validation errors attached. Do not retry when the information is simply absent from the document.

Where this scenario catches people

Batch processing appears here and the trap is the workload, not the API. 50% cheaper with up to a 24-hour window and no latency guarantee is right for an overnight run and wrong for anything a person is waiting on — and a batch request cannot run a multi-turn tool loop at all.

Read the passage once, properly

Every item in this block shares this passage, so the two minutes you spend on it are amortised across the whole block — and the details in it are load-bearing. The named tools, the stated target, the tools the agent is said to have: each of those has been the deciding fact in a question.

What is quoted, and what is argued

The passage and the primary-domain list are the exam guide's own words. Everything after them is this hive's reading of what the domains imply for this context — useful preparation, not an official statement about what the exam contains. The guide publishes 60 items and 4 scenarios per sitting and does not publish a per-scenario count, so none is quoted here.

Domain roadmap169 words

Domain 1: Agentic Architecture & Orchestration

Claude Certified Architect - Foundations (CCAR-F) › Domain 1: Agentic Architecture & Orchestration

Read full article

Domain 1: Agentic Architecture & Orchestration

Domain 1 is the biggest block on the exam, and nearly all of it reduces to one question: what is actually in control? The loop is controlled by stop_reason, not by reading the assistant's text. Ordering is controlled by a prerequisite gate, not by an instruction. A subagent's scope is controlled by the tools it was given, not by what it was told. Read every Domain 1 stem for the mechanism, and you will find that the distractors are almost always the version where a person is trusting the model to cooperate.

Domain 1 at a glance

Blueprint weight
27% of scored items
Task statements
7
Questions in this hive
66
Flashcards in this hive
48, one per published objective

What the guide measures here

  1. Design and implement agentic loops with proper termination conditions
  2. Orchestrate multi-agent systems with coordinator-subagent patterns
  3. Configure subagent invocation, context passing, and spawning
  4. Implement multi-step workflows with enforcement and handoff patterns
  5. Apply Agent SDK hooks for tool call interception and data normalization
  6. Design task decomposition strategies for complex workflows
  7. Manage session state, resumption, and forking

The order to learn them in

Working through Domain 1

  1. 1

    The agentic loop

    Start with the agentic loop. Everything else in the domain sits inside one, and the three termination anti-patterns are the most reliably examined facts here.

Where Domain 1 catches people

Stopping the loop on an iteration cap. A cap is a runaway backstop; the design is stop_reason. Assuming a subagent can see what the coordinator saw. Letting subagents talk to each other directly, which removes the observability that hub-and-spoke exists for. And decomposing so narrowly that every slice is complete and the cross-cutting question is owned by nobody.

The tell in a Domain 1 stem

When one option adds an instruction to a prompt and another changes what is structurally possible, the exam wants the structural one. Deterministic compliance — identity before a financial operation, a prerequisite before a refund — is never a prompt in this domain.

What is quoted, and what is argued

The weight and the 7 task statement titles above are quoted from the official CCAR-F exam guide v1.0. The thesis, sequence and traps are this hive's reading of that material — useful, and not the guide's own words. Asset counts are this hive's, read from the bank on 2026-08-29.

Domain roadmap157 words

Domain 2: Tool Design & MCP Integration

Claude Certified Architect - Foundations (CCAR-F) › Domain 2: Tool Design & MCP Integration

Read full article

Domain 2: Tool Design & MCP Integration

Domain 2 rests on one premise that explains most of its answers: the description is the interface. Tool selection runs on what the descriptions say, so two tools that sound alike get chosen between at close to chance, a constraint left out of a description is a constraint the model cannot respect, and a capable tool that describes itself vaguely loses to a built-in that does not. The second premise is that errors are data: an agent can only recover as well as the error told it what happened.

Domain 2 at a glance

Blueprint weight
18% of scored items
Task statements
5
Questions in this hive
46
Flashcards in this hive
43, one per published objective

What the guide measures here

  1. Design effective tool interfaces with clear descriptions and boundaries
  2. Implement structured error responses for MCP tools
  3. Distribute tools appropriately across agents and configure tool choice
  4. Integrate MCP servers into Claude Code and agent workflows
  5. Select and apply built-in tools (Read, Write, Edit, Bash, Grep, Glob) effectively

The order to learn them in

Working through Domain 2

  1. 1

    Tool descriptions

    Begin with tool descriptions — what belongs in one, and what overlap does. This is the domain's centre of gravity and the source of its most-repeated fact.

Where Domain 2 catches people

Fixing a selection problem in the system prompt when the descriptions are the cause — or the reverse, polishing descriptions when a keyword in the system prompt is overriding them. Retrying an error that was never retryable. Treating an empty result as a failure. And giving one agent every tool because it might need them, which is the specific thing that degrades selection.

Read the tool count

When a stem tells you how many tools an agent holds, that number is usually the answer. Eighteen where four or five would do is not background detail — it is the defect the question is about.

What is quoted, and what is argued

The weight and the 5 task statement titles above are quoted from the official CCAR-F exam guide v1.0. The thesis, sequence and traps are this hive's reading of that material — useful, and not the guide's own words. Asset counts are this hive's, read from the bank on 2026-08-29.

Domain roadmap150 words

Domain 3: Claude Code Configuration & Workflows

Claude Certified Architect - Foundations (CCAR-F) › Domain 3: Claude Code Configuration & Workflows

Read full article

Domain 3: Claude Code Configuration & Workflows

Domain 3 is the one that names exact paths and flags, and it expects you to know them: ~/.claude/CLAUDE.md against .claude/CLAUDE.md, .claude/commands/ against ~/.claude/commands/, .claude/rules/ with a paths field, -p, --output-format json, --json-schema, /memory, /compact. Underneath the paths sits one repeated question: who is this configuration for, and when should it load? Almost every diagnosis in this domain is scope — something is in the user-level file that should have been project-level, or loaded always when it should have loaded conditionally.

Domain 3 at a glance

Blueprint weight
20% of scored items
Task statements
6
Questions in this hive
50
Flashcards in this hive
49, one per published objective

What the guide measures here

  1. Configure CLAUDE.md files with appropriate hierarchy, scoping, and modular organization
  2. Create and configure custom slash commands and skills
  3. Apply path-specific rules for conditional convention loading
  4. Determine when to use plan mode vs direct execution
  5. Apply iterative refinement techniques for progressive improvement
  6. Integrate Claude Code into CI/CD pipelines

The order to learn them in

Working through Domain 3

  1. 1

    The CLAUDE.md hierarchy

    Learn the CLAUDE.md hierarchy cold, including which level reaches teammates through version control. The classic stem is a new team member who is not getting the instructions.

Where Domain 3 catches people

Putting team configuration in a user-scoped file, which works perfectly for its author and for nobody else. Planning a one-line fix, or executing straight into a migration that touches forty-five files. Letting the session that wrote the code review it. And running Claude Code in CI without -p, which hangs waiting for an interaction nobody is there to give.

Say the path out loud

This domain's distractors are frequently the right idea at the wrong path or with the wrong flag. If you can state the exact file and the exact option before reading the choices, most items in this domain resolve in seconds.

What is quoted, and what is argued

The weight and the 6 task statement titles above are quoted from the official CCAR-F exam guide v1.0. The thesis, sequence and traps are this hive's reading of that material — useful, and not the guide's own words. Asset counts are this hive's, read from the bank on 2026-08-29.

More Study Notes (2)

Domain 4: Prompt Engineering & Structured Output

Claude Certified Architect - Foundations (CCAR-F) › Domain 4: Prompt Engineering & Structured Output

150 words

Domain 5: Context Management & Reliability

Claude Certified Architect - Foundations (CCAR-F) › Domain 5: Context Management & Reliability

165 words

Ready to practice? Jump straight in — no sign-up needed.

Take practice tests, review flashcards, and read study notes right now.

Take a Practice Test

Claude Certified Architect - Foundations (CCAR-F) Practice Questions

Try 15 sample questions from a bank of 253. Answers and detailed explanations included.

Q1easy

Your monorepo has conventions that should only apply when someone is working on Terraform. How do .claude/rules/ files achieve conditional activation?

A.

YAML frontmatter paths fields containing glob patterns activate the rule only for matching files

B.

Rules activate only when explicitly imported with @import

C.

Filename prefixes like IF_ control loading

D.

A cron job toggles rules on and off

Show answer & explanation

Correct Answer: A

Path-specific rules use YAML frontmatter with a paths field of glob patterns - the rule loads only when Claude works with matching files. Prefix conventions, cron toggles, and import-gating are not the mechanism.

Q2medium

Your extraction service emits confidence numbers, and the review team wants to use them to prioritise its queue. What makes model confidence scores useful for routing review attention, per the guidance?

A.

Confidence scores averaged across all fields

B.

Raw confidence scores taken at face value

C.

Field-level confidence scores CALIBRATED using labeled validation sets

D.

Confidence inferred from response length

Show answer & explanation

Correct Answer: C

Calibration against labeled validation sets is what turns raw confidence into a reliable routing signal - uncalibrated scores misrank. Field-level granularity plus calibration is the guidance's pairing.

Q3easy

Your team wants every engineer Claude Code session to share the same Jira MCP server, checked into the repo. Where does the configuration go, per the exam guidance and Claude Code documentation?

A.

Project-level .mcp.json at the repository root, committed to version control

B.

The ~/.claude/settings.json hooks block

C.

A managed-policy CLAUDE.md describing how to install the server manually

D.

User-level ~/.claude.json, shared by emailing the file to teammates

Show answer & explanation

Correct Answer: A

Project-level .mcp.json at the project root is the shared-team scope: committed to version control so everyone gets the same servers (with a first-use approval step). User-level ~/.claude.json is the personal/experimental scope and is not shared. Settings hooks and prose instructions are not server configuration.

Q4medium

All failures in your pipeline surface as: search unavailable. Why does the guidance call this harmful?

A.

They slow down the coordinator's response time

B.

Generic error statuses hide valuable context from the coordinator - it cannot distinguish retryable timeouts from permanent failures or partial success

C.

Generic statuses are a security risk

D.

The phrase is too long for the context window

Show answer & explanation

Correct Answer: B

Generic statuses strip exactly the context recovery needs - what failed, whether retrying helps, what survived. The coordinator's decisions degrade to guesses; security and latency are not the cited harms.

Q5medium

How is a structured extraction typically implemented with tool use, per the guidance?

A.

Put the schema in the system prompt and hope

B.

Define the extraction tool with the JSON schema as its INPUT parameters, then extract the structured data from the tool_use response

C.

Use the schema only for post-hoc validation of prose

D.

Have the tool return free text and parse it downstream

Show answer & explanation

Correct Answer: B

The pattern: the JSON schema defines the tool's input parameters; the model fills them; the structured data is read from the tool_use block. Free-text-plus-parsing and prose validation forfeit the guarantee the mechanism exists to provide.

Q6medium

You are designing the extraction schema for a new document type. Which schema design considerations does the guidance list?

A.

Required vs optional fields, and enum fields with an other value plus a detail string for extensible categories

B.

All fields required, enums closed, no extensibility

C.

Nested schemas are prohibited

D.

Schemas must not exceed three fields

Show answer & explanation

Correct Answer: A

The guidance's schema-design considerations: choose required versus optional deliberately, and use enum-plus-other-with-detail patterns so categories can extend without schema rewrites. Closed enums and arbitrary field limits are anti-patterns it avoids.

Q7medium

Your coordinator has spent 20 turns building deep context about a research question. It spawns an analysis subagent. What context does the subagent have about the coordinator's conversation, per the exam guidance?

A.

A compressed automatic summary generated by the SDK

B.

None - subagents operate with isolated context and do not inherit the coordinator's conversation history automatically

C.

Only the last coordinator turn

D.

Full inheritance of the coordinator conversation history automatically

Show answer & explanation

Correct Answer: B

Subagents operate with isolated context - they do not automatically inherit the coordinator's conversation history (the Claude Code docs state each subagent starts a fresh, isolated context window). Whatever the subagent needs must be explicitly provided in its prompt; there is no automatic compression or last-turn shortcut.

Q8medium

Your monorepo has five packages, each with its own standards file. How does the @import syntax keep CLAUDE.md modular?

A.

@import creates a symlink between packages at build time

B.

@import executes the standards file as a shell script

C.

Each package's CLAUDE.md references its standards file with @path/to/file syntax, and the referenced file is expanded into context when loaded

D.

@import is only valid inside .claude/rules/ files

Show answer & explanation

Correct Answer: C

The @import syntax references external files (e.g., @docs/git-instructions.md) which are expanded and loaded into context alongside the referencing CLAUDE.md - the mechanism the guidance names for keeping CLAUDE.md modular per package. It neither executes code nor creates symlinks, and it works in any CLAUDE.md, not only rules files.

Q9medium

You want one conventions rule to cover every test file in the repo regardless of which directory holds it. Which pattern does the guidance use?

A.

A glob pattern in the paths field such as **/*.test.tsx applying the rule to files by type wherever they live

B.

A pre-commit hook that rewrites test files

C.

A naming convention enforced by code review only

D.

One rule per directory, manually maintained

Show answer & explanation

Correct Answer: A

Glob patterns apply conventions by TYPE across directory boundaries - **/*.test.tsx is the guide's example for all test files everywhere. Per-directory duplication is the maintenance burden globs replace; hooks and review-only enforcement are outside the rules mechanism.

Q10medium

A document-analysis subagent finds conflicting revenue values in its sources. What does the guidance have it do before reporting up?

A.

Escalate to a human analyst immediately

B.

Complete the analysis with the conflicting values INCLUDED and explicitly annotated, letting the coordinator decide how to reconcile

C.

Pick one value and note the other was wrong

D.

Omit the revenue section entirely

Show answer & explanation

Correct Answer: B

The subagent completes with conflicts included and annotated - reconciliation is the COORDINATOR's decision, made with full visibility. Picking a winner hides the conflict; omitting loses the section; premature human escalation bypasses the coordinator.

Q11easy

In the two-pass review pattern, which concerns belong in the CROSS-FILE integration pass rather than the per-file passes?

A.

A typo inside one function body

B.

An unused import within a single file

C.

A missing docstring in one file

D.

Cross-file data-flow issues - e.g., module A calls module B with arguments B no longer accepts

Show answer & explanation

Correct Answer: D

The integration pass exists for issues that only appear across file boundaries - contract mismatches, broken data flow between modules. Typos, docstrings, and unused imports are local, single-file concerns that the per-file passes own.

Q12medium

In your multi-agent research system, the coordinator's tool list omits the subagent-spawning tool entirely (Task/Agent). What happens when it tries to delegate?

A.

It spawns the subagent anyway via implicit fallback

B.

It cannot spawn any subagents - the spawning tool must be present in the agent's tools for delegation to be possible

C.

It prompts the user to approve the spawn

D.

It falls back to executing the subtask inline with reduced quality

Show answer & explanation

Correct Answer: B

The spawning tool must be included in the agent's allowed tools; omitting it removes the capability entirely - the Claude Code docs state that if you omit Agent (Task) from the tools list, the agent can't spawn any subagents. There is no implicit fallback, inline degradation, or interactive grant.

Q13easy

Your support agent's MCP tools return data the agent has to reshape before it can reason over it. In the Agent SDK, which hook pattern intercepts a tool's RESULT for transformation BEFORE the model processes it?

A.

PostToolUse - runs after the tool succeeds and can transform what the model receives

B.

SessionEnd - fires at session teardown

C.

UserPromptSubmit - fires when the user sends a message

D.

Stop - fires when the model finishes its turn

Show answer & explanation

Correct Answer: A

PostToolUse is the hook that intercepts tool results after execution, enabling transformation before the model sees them - the guide's named example for data normalization. UserPromptSubmit, SessionEnd, and Stop fire at unrelated lifecycle points.

Q14medium

Documents arrive whose type is unknown - invoices, contracts, or receipts - each with its own extraction schema. How do you guarantee structured output regardless of which applies?

A.

Force one fixed schema for all document types

B.

Ask the model to state which schema it will use in prose first

C.

Set tool_choice: any so the model MUST call a tool, choosing among the extraction schemas

D.

Set tool_choice: auto and retry when text comes back

Show answer & explanation

Correct Answer: C

With multiple extraction schemas and unknown document type, tool_choice any guarantees A tool call - structured output - while letting the model pick the right schema. auto permits text escapes; a single forced schema misfits most documents; prose declarations are not tool calls.

Q15medium

lookup_customer returns three records matching the name the customer gave. What does the guidance instruct?

A.

Pick the record with the highest credit limit

B.

Pick the most recently created record

C.

Merge all three records into one profile

D.

Request additional identifiers to clarify which customer is correct, rather than selecting based on heuristics

Show answer & explanation

Correct Answer: D

Multiple matches demand clarification - ask for additional identifiers - never heuristic selection on recency, value, or guesswork. Merging records is a data-integrity violation on top of the misidentification risk.

These are 15 of 253 questions available. Take a practice test →

Claude Certified Architect - Foundations (CCAR-F) Flashcards

240 flashcards for spaced-repetition study. Showing 30 sample cards below.

Agentic loops(6 cards shown)

Question

What drives each turn of the agentic loop?

Answer

stop_reason. A value of tool_use means Claude wants tools run: execute them, append the results, send again. end_turn means it has finished. The loop is a while keyed on that field — never on anything in the assistant's text.

Question

Where do tool results go between iterations?

Answer

Into the conversation history, as a tool_result message alongside the assistant message that requested it. The API is stateless between calls, so the history you send is the only thing carrying the result forward.

Question

Model-driven control flow versus a pre-configured decision tree — what actually differs?

Answer

A decision tree routes on surface keywords and cannot adapt once it has branched. In an agentic loop Claude picks the next tool from the full context, including what earlier tools returned, which is what lets it handle an ambiguous request.

Question

Sketch the agentic loop's control flow.

Answer

While stop_reason is tool_use: run the requested tools, append their results to the history, and call again. Break when it is end_turn. No turn counters, no keyword matching.

Question

Your agent repeats a mistake the validator already caught. What did the loop fail to do?

Answer

Add the tool results to the context between iterations. Without them the next request is identical to the last one, so the model has nothing new to reason from and makes the same call again.

Question

Name the three loop-termination anti-patterns.

Answer

Parsing natural-language text for phrases like 'I have completed your request'; using an arbitrary iteration cap as the PRIMARY stopping mechanism; and treating any non-empty assistant text as completion. A cap is a runaway backstop, not the design.

Batch processing strategy(8 cards shown)

Question

State the Message Batches API's three headline terms.

Answer

50% cost savings, a processing window of up to 24 hours, and no guaranteed latency SLA. The saving is paid for in time, and the absent SLA is the part that decides suitability.

Question

Which workloads belong in a batch, and which must never be?

Answer

Non-blocking, latency-tolerant work: overnight reports, weekly audits, nightly test generation. Never a blocking workflow such as a pre-merge check — nobody can wait up to 24 hours to merge.

Question

Can a batch request run an agentic tool loop?

Answer

No. The batch API does not support multi-turn tool calling within a single request — it cannot execute a tool mid-request and feed the result back. Work needing a loop belongs on the synchronous API.

Question

How do you know which batch response answers which request?

Answer

The custom_id you set on each request comes back on its response. Without it, correlating a returned result to its input is guesswork.

Question

Match the API to the workflow: a pre-merge check, and a weekly analysis.

Answer

Synchronous for the pre-merge check — it blocks a person. Batch for the weekly analysis — nothing waits on it, so the 50% saving costs nothing that matters.

Question

You owe a 30-hour SLA and batches take up to 24. How often do you submit?

Answer

Every 4 hours. Worst case a document waits the full 4-hour window before submission and then the full 24-hour processing, which is 28 — inside 30. Submission frequency is what makes an SLA survive an unguaranteed processing window.

Question

Part of a batch failed. What do you resubmit?

Answer

Only the failed items, identified by custom_id, with whatever modification the failure calls for — chunking the documents that exceeded context limits, for instance. Resubmitting the whole batch pays twice for everything that already worked.

Question

What do you do before sending 10,000 documents through a batch?

Answer

Refine the prompt on a sample. First-pass success is what a batch's economics rest on — a flaw found on document one is found on all 10,000, and every resubmission cycle spends the saving again.

Built-in tools: Read, Write, Edit, Bash, Grep, Glob(9 cards shown)

Question

Which tool searches inside files?

Answer

Grep — file CONTENTS, for patterns like function names, error messages or import statements. The question it answers is "where does this text appear", not "which files exist".

Question

Which tool matches file paths?

Answer

Glob — file NAMES and extensions. It answers "which files exist by this pattern" and never looks at what is inside them.

Question

Read/Write versus Edit — what is each for?

Answer

Read and Write handle the whole file. Edit makes a targeted modification by matching unique text, so it changes one place without rewriting everything around it.

Question

Edit fails because the anchor text appears more than once. What is the fallback?

Answer

Read the file, then Write it back with the change. Retrying Edit with the same non-unique anchor fails identically — the fallback is the reliable path, not a workaround for a broken tool.

Question

You need every caller of a function across the codebase. Which tool?

Answer

Grep. Callers are content, and the same goes for tracking down where an error message is produced — both are searches through what files say rather than what they are named.

Question

You need every test file under a directory tree. Which tool?

Answer

Glob, with a pattern like **/*.test.tsx. It is a naming question, so it never opens a file.

Question

There is no unique anchor for the change you need. Walk through the alternative.

Answer

Read the full contents, apply the change to what you have, and Write the result. You are supplying the whole file rather than asking the tool to locate a spot that cannot be located.

Question

How do you build understanding of an unfamiliar codebase?

Answer

Incrementally: Grep for entry points, then Read to follow the imports and trace the flows that matter. Reading everything up front fills the context with files the task never touches and dilutes attention on the ones it does.

Question

A function is re-exported through wrapper modules. How do you find its real usage?

Answer

Identify all the exported names first, then search for each name across the codebase. Searching only the original name finds the wrappers and misses every call that goes through them.

Claude Code in CI/CD(7 cards shown)

Question

Which flag runs Claude Code non-interactively?

Answer

-p, also spelled --print. It is what makes the CLI usable inside an automated pipeline, where there is nobody to answer a prompt.

Question

Which two flags enforce structured output in CI?

Answer

--output-format json and --json-schema. The first makes the output machine-readable; the second makes it conform to a shape the pipeline can rely on.

Question

How does a CI-invoked run learn your project's conventions?

Answer

From CLAUDE.md — testing standards, fixture conventions, review criteria. CI has no conversation history to draw on, so anything not written down is unavailable to it.

Question

Why not have the session that wrote the code review it?

Answer

It is less effective at reviewing its own changes. It carries the reasoning that produced them, so the assumptions behind a mistake are shared rather than examined. An independent instance sees the diff without the story.

Question

Your CI job hangs waiting for input. What was missing?

Answer

The -p flag. Without it the run expects interaction, and a pipeline with nobody at the keyboard waits until it times out.

Question

You want findings posted as inline PR comments. What produces them?

Answer

--output-format json with --json-schema, giving machine-parseable structured findings the pipeline can place on specific lines. Prose has to be scraped, and scraping is where line numbers get lost.

Question

Your re-run after new commits repeats every comment from the last review. What is missing?

Answer

The prior findings in context, plus an instruction to report only new or still-unaddressed issues. Each run is independent, so without the earlier review it rediscovers everything and posts it again.

Showing 30 of 240 flashcards. Study all flashcards →

Related Study Resources

Explore other free certification prep and study materials on BrainyBee.

AWS Certified Cloud Practitioner (CLF-C02)

854 questions · 163 notes

AWS Certified Solutions Architect - Associate (SAA-C03)

1408 questions · 204 notes

AWS Certified Machine Learning Engineer - Associate (MLA-C01)

724 questions · 160 notes

AWS Certified CloudOps Engineer - Associate (SOA-C03)

840 questions · 148 notes

AWS Certified Security - Specialty (SCS-C03)

980 questions · 130 notes

Microsoft Azure Fundamentals (AZ-900)

680 questions · 96 notes

AWS Certified Advanced Networking - Specialty (ANS-C01)

1156 questions · 231 notes

Microsoft Azure AI Fundamentals (AI-901)

663 questions · 89 notes

Ready to ace Claude Certified Architect - Foundations (CCAR-F)?

Access all 253 practice questions, 4 timed mock exams, study notes, and flashcards — no sign-up required.

Start Studying — Free
Explore All HivesBlogHome

© 2026 BrainyBee. Free AI-powered exam prep.