CCAR-F
Free Claude Certified Architect - Foundations Practice Test
12 exam-style questions with full explanations — no sign-up. Score yourself, then close your gaps with the full course.
Exam fee ~$1255 exam domainsLevel Advanced2 timed practice tests in the course
✅ Free practice — no sign-up📝 Real exam-style questions💡 Detailed explanations💸 30-day money-back via Udemy
An architect is designing a multi-agent research system where a coordinator agent delegates five parallel research tasks to subagents. The coordinator's conversation history is long and contains unrelated prior tool calls. To prevent subagents from being confused by irrelevant history while still giving them what they need to complete their task, which design should the architect use?
Subagents in Claude's architecture run with their own isolated context windows; passing only the task-specific instructions and necessary context avoids context pollution and keeps each subagent focused, which is a core principle of coordinator-subagent design.
A monorepo has a root CLAUDE.md with general coding standards and a backend/CLAUDE.md with database-specific conventions. When an architect asks Claude Code to modify a file inside backend/, which behavior BEST describes how these files are used?
CLAUDE.md is loaded hierarchically from the user home directory, project root, and nested subdirectories. All applicable files are loaded additively into context, with directory-level files layering more specific guidance on top of the broader project-level instructions.
A team needs to extract structured invoice data (vendor, amount, due date) from scanned PDFs and insert it directly into a downstream database without any human review step. The output MUST always conform to a strict JSON schema or the pipeline will crash. Which approach BEST guarantees schema conformance?
Forcing a specific tool via tool_choice guarantees Claude's response is the tool_use block with arguments constrained by the tool's input schema, producing schema-conformant structured output reliably at the API level rather than relying on the model to follow free-text formatting instructions.
A company wants to give Claude the ability to query its internal ticketing system, and wants this exact integration reused across Claude Desktop, Claude Code, and a custom internal agent application. Which approach BEST meets these requirements?
MCP servers exist specifically to provide a single, reusable integration that any MCP-compatible client (host) can connect to via a standardized client-server connection, avoiding duplicated per-app implementations.
An agent has been running a multi-hour coding session and its context window is approaching the limit. The architecture needs to continue the session without losing critical task state. Which approach BEST maintains reliability across the context boundary?
Structured checkpointing (task state, decisions, file/tool state) combined with compaction of older turns preserves the information the agent needs to continue reliably while keeping the active context manageable. This is the standard pattern for long-running agent sessions.
A team wants to guarantee that a validation script runs automatically every time the agent writes to a file, without relying on the model choosing to call it via the system prompt. Which mechanism should they implement?
Hooks provide deterministic, code-enforced checkpoints tied to specific lifecycle events (such as after a tool call), guaranteeing an action runs regardless of model behavior—unlike prompt-based instructions which are probabilistic.
An architect wants Claude to automatically apply a specialized capability (with its own instructions and bundled scripts) whenever the task context matches a particular kind of work, without the user needing to remember and type a specific trigger. Which Claude Code feature is the BEST fit?
Skills are discoverable capabilities with a description that Claude evaluates to decide when relevant, and they can bundle instructions plus supporting scripts/resources. They are model-invoked based on contextual fit, matching this requirement exactly.
An architect is designing a nightly pipeline that summarizes 80,000 customer support transcripts. There is no requirement for a response within minutes, and the team wants to minimize API cost. Which processing approach is MOST appropriate?
The Message Batches API is purpose-built for large-volume, non-realtime workloads, offering a 50% cost discount compared to standard API pricing, with results typically available within 24 hours, which fits this nightly, cost-sensitive use case exactly.
An architect needs Claude to execute Python code for data analysis directly within a conversation, with minimal setup overhead. Which option BEST satisfies this requirement?
Anthropic provides a built-in, Anthropic-hosted code execution tool that can be enabled directly through the API, eliminating the need to build or maintain custom sandboxing infrastructure.
During a customer-support agent workflow, the agent receives an instruction that could reasonably be interpreted two different ways, and acting on the wrong interpretation would cause irreversible harm (e.g., issuing a refund to the wrong account). What is the BEST practice for handling this?
When an action is irreversible and ambiguity exists, the reliable pattern is to escalate — ask for clarification or hand off to a human — before acting, rather than guessing.
An agent session running a long multi-step workflow is interrupted (e.g., process restart). An architect wants to resume execution exactly where it left off, preserving all prior tool calls and reasoning state. What must be used to resume correctly?
Session resumption relies on referencing the original session ID so the stored conversation/tool-call history can be reloaded and execution continued from that exact state.
An architect wants Claude Code to run `git diff` and `git log` commands without prompting for confirmation each time, while still requiring approval for all other Bash commands. Which settings.json permissions configuration achieves this?
The allow list accepts tool-specific patterns like Bash(<command-prefix>:*), which pre-approves exactly those matching commands while leaving all other Bash invocations subject to the default confirmation prompt.
Ready for the real thing?
The full course has two full-length practice tests, video lessons for every exam domain, hands-on labs and detailed answer explanations.
Start my full course on Udemy →CCAR-F exam — quick answers
How much does the CCAR-F exam cost?
The exam fee is approximately $125 and varies by region — confirm current pricing with the certification vendor before you book.
What topics are on the exam?
It covers 5 domains: Agentic Architecture and Orchestration (27%), Claude Code Configuration and Workflows (20%), Prompt Engineering and Structured Output (20%), Tool Design and MCP Integration (18%), Context Management and Reliability (15%). The full course has a dedicated chapter, lab and practice-test coverage for each.
Is this practice test really free?
Yes — all questions on this page are free with explanations and no sign-up. The paid Udemy course adds two full-length timed exams, video lessons and hands-on labs.
Will this prepare me for the real exam?
The questions mirror the real exam's style and are mapped to the official domains. This is exam-focused preparation — combine the free test with the full course's timed simulations to gauge your readiness.