CCAR-F Exam Flashcards

Q: What does CCAR-F focus on?

CCAR-F prep should emphasize Claude architecture choices, prompt and tool design, Claude Code workflows, MCP integrations, context management, evaluation, and production reliability trade-offs.

Q: What is the CCAR-F retake schedule?

Wait 14 days after the first failed attempt, 30 days after the second, and 90 days after the third; the limit is four attempts per rolling 12 months.

Question 1

CCAR-F architectural default

Accepted Answer

Start with the simplest design that meets the requirement, then add agentic complexity only when the simpler approach fails measurable tests.

Question 2

Single LLM call with retrieval

Accepted Answer

A strong baseline for many applications: retrieve relevant context, provide clear instructions and examples, and evaluate before adding loops or agents.

Question 3

Agent loop

Accepted Answer

A model repeatedly plans, calls tools, observes results, and decides whether to continue. Use it when the path cannot be fully scripted.

Question 4

Prompt chaining

Accepted Answer

Break a task into ordered steps where each model call feeds the next. It improves control when the workflow sequence is predictable.

Question 5

Routing pattern

Accepted Answer

Classify the input, then send it to the best specialized prompt, tool, model, or workflow. Use when inputs fall into known categories.

Question 6

Parallelization pattern

Accepted Answer

Run independent subtasks at the same time and combine results. It helps with speed or diverse checks when subtasks do not depend on each other.

Question 7

Orchestrator-worker pattern

Accepted Answer

One model decomposes work and delegates to worker calls or agents. Use when subtasks are dynamic and cannot be enumerated ahead of time.

Question 8

Evaluator-optimizer pattern

Accepted Answer

One call produces an output and another evaluates it against criteria, creating feedback for revision. Useful when quality can be judged clearly.

Question 9

Subagent design

Accepted Answer

Give each subagent a narrow role, clear tool access, and a bounded output contract. Broad overlapping roles create coordination failures.

Question 10

Model selection trade-off

Accepted Answer

Choose model capability based on task risk, ambiguity, latency, and cost. Use stronger models for complex reasoning and smaller models for routine steps.

Question 11

Tool-use schema

Accepted Answer

A tool schema should name the action, define required inputs, constrain types, and make invalid calls hard for the model to produce.

Question 12

Tool description quality

Accepted Answer

The model relies on the tool name and description to decide when to call it. Ambiguous descriptions lead to wrong tool selection.

Question 13

Tool error handling

Accepted Answer

Return concise, structured errors that explain what failed and whether retry, correction, or escalation is appropriate.

Question 14

Least-privilege tools

Accepted Answer

Expose only the operations and data a workflow needs. Tool access should be scoped by role, environment, and user authorization.

Question 15

MCP purpose

Accepted Answer

Model Context Protocol standardizes how Claude connects to external tools, resources, and prompts through MCP servers.

Question 16

MCP tools

Accepted Answer

Tools are callable actions exposed by an MCP server, such as querying a system, creating an object, or running a bounded operation.

Question 17

MCP resources

Accepted Answer

Resources expose contextual data that a model can read, such as files, records, schemas, or documentation snippets.

Question 18

MCP prompts

Accepted Answer

Prompts are reusable prompt templates or workflows provided by an MCP server to guide common tasks.

Question 19

stdio MCP transport

Accepted Answer

stdio is commonly used for local MCP servers launched as subprocesses. It is simple for local developer workflows.

Question 20

HTTP MCP transport

Accepted Answer

HTTP-based transport is better suited for remote services, hosted integrations, and network-accessible MCP servers.

Question 21

Claude Code project memory

Accepted Answer

CLAUDE.md captures durable project instructions, conventions, and constraints that Claude Code should follow across tasks.

Question 22

Claude Code settings

Accepted Answer

settings.json controls local configuration such as permissions, tool behavior, and workflow defaults depending on the project setup.

Question 23

Claude Code permissions

Accepted Answer

Permissions determine which operations can run automatically and which require confirmation. Use stricter permissions for risky write or shell actions.

Question 24

Slash commands

Accepted Answer

Slash commands package repeatable workflows so developers can trigger common project tasks with consistent instructions.

Question 25

Hooks

Accepted Answer

Hooks run configured actions at workflow points, such as validation, formatting, logging, or policy checks.

Question 26

Headless Claude Code usage

Accepted Answer

Headless or CI usage requires explicit prompts, predictable outputs, scoped permissions, and validation because no human is steering interactively.

Question 27

System prompt role

Accepted Answer

The system prompt sets durable role, constraints, output expectations, and safety boundaries for the model.

Question 28

Task framing

Accepted Answer

State the goal, inputs, constraints, and success criteria. Missing criteria force the model to infer requirements.

Question 29

Examples in prompts

Accepted Answer

Good examples show the model the desired style, format, level of detail, and edge-case handling.

Question 30

Structured output

Accepted Answer

Structured output asks the model to produce a predictable shape, often enforced by a schema or tool call for reliable parsing.

Question 31

JSON schema for output

Accepted Answer

A schema should define required fields, types, enums, and nesting. This reduces parser failures and ambiguous responses.

Question 32

Stop sequences

Accepted Answer

Stop sequences end generation at predictable boundaries. They help prevent trailing text when exact parsing matters.

Question 33

Extended thinking fit

Accepted Answer

Use deeper reasoning for complex planning or trade-off decisions, not for simple extraction or deterministic formatting.

Question 34

Context window risk

Accepted Answer

Long contexts can bury important instructions or exceed limits. Prioritize, summarize, and retrieve only the context needed for the task.

Question 35

Prompt caching

Accepted Answer

Prompt caching can reduce cost and latency when stable, repeated context is reused across requests.

Question 36

RAG purpose

Accepted Answer

Retrieval-augmented generation supplies relevant external context at request time instead of relying only on the model's stored knowledge.

Question 37

Chunking for retrieval

Accepted Answer

Chunks should be large enough to preserve meaning but small enough to retrieve precisely. Poor chunking causes missing or noisy context.

Question 38

Grounded answer behavior

Accepted Answer

A grounded workflow should cite or use retrieved context and refuse to invent details when the context is insufficient.

Question 39

Retries

Accepted Answer

Retries help transient failures but should use limits, backoff, and idempotency safeguards to avoid repeated harmful actions.

Question 40

Timeouts

Accepted Answer

Timeouts prevent workflows from hanging indefinitely. Set them based on expected tool latency and user experience requirements.

Question 41

Evals

Accepted Answer

Evals measure whether the system performs the intended task reliably. They should cover realistic success cases and known failure modes.

Question 42

Guardrails

Accepted Answer

Guardrails constrain unsafe, invalid, or out-of-scope behavior through prompts, tools, policies, validation, or human review.

Question 43

Human-in-the-loop

Accepted Answer

Use human approval for high-impact actions, irreversible changes, sensitive data access, or uncertain model judgments.

Question 44

Latency optimization

Accepted Answer

Reduce latency by using smaller models where appropriate, caching, parallel independent calls, and limiting unnecessary context.

Question 45

Cost optimization

Accepted Answer

Control cost by routing simple tasks to cheaper models, trimming context, caching stable prompts, and avoiding needless agent loops.

Question 46

Production observability

Accepted Answer

Log inputs, tool calls, outputs, latency, cost, and errors at the right level so failures can be diagnosed without leaking sensitive data.

Question 47

Prompt injection defense

Accepted Answer

Treat retrieved or user-provided content as data, not instructions. Keep trusted system instructions separate and validate tool actions.

Question 48

Sensitive data handling

Accepted Answer

Limit what data enters prompts, redact where possible, and scope tools so the model cannot access data unrelated to the task.

Question 49

Architecture decision record

Accepted Answer

Document why a model, tool, prompt pattern, or agent design was chosen so future teams can evaluate trade-offs and changes.

Question 50

CCAR-F exam mindset

Accepted Answer

Prefer answers that are simple, testable, least-privilege, observable, and reliable under failure conditions.

Free CCAR-F Exam Flashcards

Filter by Topic

Jump to Card

About These CCAR-F Flashcards

Topics Covered

Frequently Asked Questions

Claude Certified Architect - Foundations (CCAR-F; formerly CCA-F)

Explore More Anthropic Claude Certifications

More From This Family