Try one now
An autonomous data-cleaning agent on Claude Sonnet 5 works through batches of financial ledgers. On a row with a malformed date it calls parse_date with the same arguments again and again. Every one of those calls returns in about 40 ms with HTTP 200 and the same unparseable payload, so nothing in the transport layer looks wrong, and the loop runs until the task exhausts its tokens and the rest of the batch is left unprocessed. Operations needs the agent to notice the dead end while it is happening, change approach on that row, and carry on through the remaining ledgers.
Which control stops the repetition while the run is still usable?
Answer it and read the explanation →About the CCAR-F exam
- Length
- 120 minutes
- Questions
- 60 items
- Passing score
- 720 on a 1,000 scale
- Price
- $125 USD
- Question types
- Multiple choice and multiple response
Checked against the provider's certification page and exam guide on 2026-09-15. Providers change fees and formats; confirm before you book. Register for the official exam ↗
What the exam covers
120 minutes, 60 questions, and a 720 out of 1,000 passing score. Anthropic lists multiple-choice and multiple-response items.
- Agentic architecture and orchestration27%
- Tool design and MCP integration18%
- Claude Code configuration and workflows20%
- Prompt engineering and structured output20%
- Context management and reliability15%
Weights are from the provider's published exam guide.
Who this is for. Architects building reliable applications and agents with Claude.
How to study. Reason about where an agent should hand control to deterministic code, request a tool, or stop for human review.
Our bank has 59 questions and uses single-answer multiple choice. Official exams may use additional formats. This is an independent practice set.
All 59 practice questions
Open a question to try it and read the reasoning. Related questions may also appear in other relevant tracks.
Agentic Architecture & Orchestration 16
- Which control stops the repetition while the run is still usable?medium
- Which control flow governs the loop?easy
- What most reliably enforces the ticket requirement?medium
- Which architecture best fixes the misrouting?medium
- Which design is most appropriate?medium
- Where should the system normalize the values?medium
- What should the coordinator pass to the final writer?medium
- Which change cuts the waiting without adding overhead?easy
- Which offering should the team build on?easy
- What is the most effective first adjustment?medium
- How should effort be set across the orchestrator and its subagents?easy
- Which memory design stops the repeated work?medium
- What is the correct way for the loop to handle this response?medium
- Which change cuts the round trips and keeps the raw payloads out of context?medium
- Which configuration meets both goals?hard
- What change makes the requests valid again?medium
Claude Code Configuration & Workflows 12
- How should the run be configured?medium
- Which configuration split matches those requirements?medium
- Which flags produce a result the checker can parse?medium
- Which configuration split fits both needs?medium
- Which hook event runs the check?easy
- How should the team make Claude Code follow those conventions?easy
- Which restructuring restores adherence?medium
- What explains the blocked commands?hard
- Which command starts the second session?easy
- Which rule of the permission system explains the block?hard
- Which frontmatter change fits this skill?medium
- How should the reviewer be defined?medium
Context Management & Reliability 8
- Which compaction strategy relieves the pressure without losing those facts?hard
- Which architecture keeps cost and latency down without losing quality on the hard cases?medium
- Which configuration keeps the window manageable without losing the IDs?medium
- What should the assistant hand to the reviewer?medium
- Which fact must survive compaction?medium
- How should the later windows be started?medium
- What is the correct way to handle these responses?medium
- What must the backend change?medium
Prompt Engineering & Structured Output 12
- Which change to the request restores that control?medium
- What explains the zero hit rate?hard
- Which change should the team make first?medium
- Which approach fits this workload best?medium
- Which post-parse step addresses those rejections?hard
- Which output contract fits that consumer?medium
- Which check does the application still have to run?medium
- How should the examples be revised?medium
- Which prompt restructuring should you try first?medium
- What should replace the prefill?medium
- Which change makes the model itself emit an integer?medium
- Which revision makes the schema valid without losing the bounds?hard
Tool Design & MCP Integration 11
- How should the tool return a result of that size?medium
- Which transport fits that requirement?medium
- How should the server represent that failure?medium
- Which change to the tool surface fixes this?medium
- What should the tool result carry instead?medium
- Where should the server definition live?medium
- Which change to the tool surface fixes the routing?hard
- Which change keeps a successful injection from doing damage?hard
- Which client behaviour should the host change?medium
- Which result should the tool return to collect them?medium
- Where should the diagnostic lines go?medium
Common questions
What can I practice for Claude Certified Architect: Foundations?
59 independent single-answer practice questions covering selected topics: Agentic architecture and orchestration, Tool design and MCP integration, Claude Code configuration and workflows, Prompt engineering and structured output, Context management and reliability. Each includes an explanation and source link.
Is this an official certification or exam dump?
No. AI CertArena writes independent practice scenarios. We are not affiliated with or endorsed by the exam providers, and we do not publish live exam questions.
Does my practice score predict an exam pass?
No. Your score is the percentage correct in this practice set, including skipped questions. Our 80% practice goal is a study target, not an official cut score or calibrated pass prediction.
How should I use this practice bank?
Try a short set, review the explanation for every miss, and repeat those questions later. Use the official exam guide to check topics that this starter bank does not yet cover.
Can I save and share my progress?
Sign in with Google to keep completed attempts and topic evaluations across devices. You can sign in after finishing a set and it still counts. You can also choose to publish a saved practice result and download an image to share. A practice badge is not a vendor credential.