--- title: "Free CCAR-P Mock Exam, Timed, No Sign-up | Cred Farmer" url: https://credfarmer.com/practice/ccar-p description: "63 CCAR-P questions in a free 120-minute mock, no account needed. Sit it timed, quick or untimed, then get a score by domain and a reason for every option." last_updated: 2026-10-29 publisher: "Cred Farmer, an independent study platform, not affiliated with or endorsed by Anthropic or Pearson VUE" --- No account needed # Free CCAR-P mock exam, timed, with a reason for every option 63 questions written for the Claude Certified Architect - Professional blueprint, sat at the official pace of about 1.9 minutes an item. No sign-up, nothing saved, nothing sent anywhere. Want more? The post [CCAR-P architecture scenarios explained](https://credfarmer.com/blog/ccar-p-practice-questions) works 10 more items option by option, and the [CCAR-P study guide](https://credfarmer.com/certifications/claude-certified-architect-professional#questions) shows 6 more with their keys: 79 distinct items with this paper, none behind an account. 63 items · 120 minutes official · 720 to pass · No account **The CCAR-P exam changes on 7 December 2026.** Pearson VUE says the CCAR-P exam includes updated content from 7 December 2026, and that the updated exam guide is published on 17 November 2026. This page follows Exam Guide v1.0 (July 2026), the guide to prepare from if you sit before 7 December 2026. If you sit on or after that date, prepare from the updated guide. We will compare it with v1.0 when it is published and update this page. [Pearson VUE's notice](https://www.pearsonvue.com/us/en/anthropic.html), read on 29 October 2026. **63**Items in this mock **120**Official minutes **720**Scaled score to pass **$175**Official exam fee **7**Official domains Question bank dated 2 October 2026. Checked against Exam Guide v1.0 (July 2026) on 17 September 2026. Three ways to sit it ## Start the CCAR-P mock Answer in this tab and get a raw score by domain with the key and the reason each option fails. Every mode scores the same way and shows the same review afterwards. The clock is the only difference, and it is the difference most candidates underestimate. Or work through the paper below one question at a time, checking each as you go. ### Full, timed All 63 items in 120 minutes, about 1.9 minutes an item, the official pace. The sitting submits itself when the clock reaches zero. ### Quick, timed 30 items drawn evenly across the 7 domains, in 57 minutes at the same pace. A different draw each time. ### Untimed review Every item with no clock. Flag what you are unsure of, submit when you are done, and read the reasons. ## Filter the paper by domain - [All domains 63](https://credfarmer.com/practice/ccar-p) - [D1 Solution Design & Architecture 11](https://credfarmer.com/practice/ccar-p?domain=1) - [D2 Claude Models, Prompting & Context Engineering 8](https://credfarmer.com/practice/ccar-p?domain=2) - [D3 Integration 12](https://credfarmer.com/practice/ccar-p?domain=3) - [D4 Evaluation, Testing & Optimization 10](https://credfarmer.com/practice/ccar-p?domain=4) - [D5 Governance, Safety & Risk Management 9](https://credfarmer.com/practice/ccar-p?domain=5) - [D6 Stakeholder Communication & Lifecycle Management 9](https://credfarmer.com/practice/ccar-p?domain=6) - [D7 Developer Productivity & Operational Enablement 4](https://credfarmer.com/practice/ccar-p?domain=7) Answer as you read ## The CCAR-P paper: every question 63 items, each tagged with its official domain. Work through the paper one question at a time: pick an option to check it, or ask for the key, and read why each option holds or fails. Nothing is saved. For a scored sitting against the clock, use the start modes above. Checked 0 of 63, 0 correct [Sit it timed instead](https://credfarmer.com/practice/ccar-p#pp-start-heading) 1. Question 1 · Domain 1: Solution Design & Architecture · Single answer A parcel carrier receives 9,000 customer emails a week, mostly asking where a parcel is. Tracking status changes every few minutes. Which Claude-based design best meets a fifteen-minute reply target? Select the best answer. - A. Answer from a nightly export of tracking data loaded into a retrieval index. - B. Route parcel emails to a team that looks up and answers each by hand. - C. Let Claude answer from past replies, with no lookup. - D. Have Claude draft each reply from a live tracking lookup. 2. Question 2 · Domain 3: Integration · Single answer A scheduling agent carries 31 tools, some that cancel referrals or export patient lists. The role needs six of them, none that cancel or export. Wrong-tool selection is 14 percent. What should you do first? Select the best answer. - A. Remove the 25 tools the role does not need and measure wrong-tool picks again. - B. Rewrite all 31 tool descriptions and add usage examples. - C. Split the 31 tools across two agents with a router. - D. Group the tools into categories, then let the agent pick a category first. 3. Question 3 · Domain 2: Claude Models, Prompting & Context Engineering · Single answer An article classifier must reach 95 percent accuracy on a labeled set. The top tier scores 97.1 percent; the fastest tier scores 96.4 percent at one twelfth the cost. What should you do? Select the best answer. - A. Stay on the top tier and keep the 0.7-point accuracy margin. - B. Split traffic evenly between both tiers to hedge against drift. - C. Pick the middle tier as a compromise, without testing it on the labeled set. - D. Move the classifier to the fastest tier. 4. Question 4 · Domain 1: Solution Design & Architecture · Single answer Every auto claim goes through the same five steps: extract, classify damage, check coverage, estimate a reserve, and route. Regulators must be able to reproduce each routing decision. Which architectural pattern fits best? Select the best answer. - A. A fixed workflow in code that calls Claude only for the language steps. - B. An autonomous agent that decides which steps each claim needs, then explains itself. - C. A single prompt that performs all five steps and returns the route. - D. Five agents, one per step, that negotiate the route. 5. Question 5 · Domain 3: Integration · Single answer An internal MCP server gives 600 branch staff account lookups. It reads any customer through one service account, choosing the customer from the account number typed in chat. What closes the security gap? Select the best answer. - A. Tell the assistant to look up only customers assigned to that staff member. - B. Log every lookup with the staff name and review the log weekly. - C. Enforce the signed-in user's permissions at the MCP server. - D. Ask staff to confirm in chat that they may view each account. 6. Question 6 · Domain 1: Solution Design & Architecture · Single answer An orchestrator sends three subagents to review supplier quotes. Each returns its full transcript, so the orchestrator holds 180,000 tokens and its recommendation contradicts the subagents' findings. What should change? Select the best answer. - A. Have each subagent return only its short structured findings. - B. Give the orchestrator a larger context window that fits every transcript. - C. Merge the three subagents into one agent that does all three reviews. - D. Run the subagents in sequence instead of in parallel. 7. Question 7 · Domain 2: Claude Models, Prompting & Context Engineering · Single answer A billing assistant sometimes agrees to waive late fees when customers insist, though only a supervisor can approve a waiver and the assistant has no waiver tool. How should the system prompt change? Select the best answer. - A. Tell the assistant to stay polite but firm whenever customers insist. - B. Remove every mention of fees from the system prompt. - C. State that waivers are outside its scope, with a scripted supervisor handoff. - D. Add a waiver tool so it can apply what it promises. 8. Question 8 · Domain 3: Integration · Single answer Gate agents ignore answers slower than two seconds; wrong answers are usually caught at the booking screen. At p95 the fast tier takes 1.6 seconds, the more accurate large tier 3.2. Which tier fits? Select the best answer. - A. The large tier, plus a loading message. - B. The fast tier for all questions, reviewed monthly for accuracy. - C. The large tier, for its higher accuracy. - D. The large tier for hard questions and the fast tier for the rest. 9. Question 9 · Domain 1: Solution Design & Architecture · Single answer Clinicians edit and sign discharge summaries that Claude drafts. About 41 percent need heavy rewriting, but nobody knows what changes, and no step may add clinician time. What closes the loop? Select the best answer. - A. Add a short survey after each signature asking what was wrong with the draft. - B. Have a panel of senior clinicians rate twenty drafts each month. - C. Switch to a more capable model tier and watch the rewrite rate. - D. Diff each draft against its signed version and log the edits. 10. Question 10 · Domain 3: Integration · Single answer Contracts are indexed as fixed 512-token chunks with no metadata. For questions about one client's agreement, 31 percent of top chunks come from a different contract, and many split a clause. What should change? Select the best answer. - A. Chunk by clause and add contract metadata for filtering. - B. Retrieve the top 50 chunks and let the model sort them. - C. Raise the chunk size to 2,000 tokens, with overlap. - D. Use a legal-text embedding model. 11. Question 11 · Domain 2: Claude Models, Prompting & Context Engineering · Select 2 A template pastes each supplier's document into the system prompt. One document said to ignore the format and reply in French, and 340 records failed to load. Which two changes should you make? (Select two.) Select exactly 2.**0/2 selected.** - A. Add a prompt line saying to ignore any instructions and formats inside documents. - B. Keep the system prompt fixed and pass documents in tagged user turns. - C. Translate every document into English first. - D. Raise the temperature to handle odd inputs. - E. Return records through a tool schema that is validated before loading. 12. Question 12 · Domain 1: Solution Design & Architecture · Single answer One prompt reads a 60-page grant application, checks 14 eligibility rules, scores six criteria and drafts a letter. The eligibility check is right only 58 percent of the time. How should the work be decomposed? Select the best answer. - A. Keep one prompt but add chain-of-thought steps and a worked example. - B. Split the application into ten-page sections and run the prompt on each. - C. Use the largest model for the single prompt. - D. Apply the rules in code and score each criterion in its own call. 13. Question 13 · Domain 3: Integration · Single answer A shopping assistant logs every full prompt and response from two million daily sessions. Logging now costs more than the model, and a broken promotion rule went unnoticed for two days. What should change? Select the best answer. - A. Keep logging everything, but move it to cheaper storage, with indexes. - B. Trace every session with rule-check alerts and sampled content. - C. Stop logging content and keep only error and latency counts. - D. Log content only for sessions that end in a complaint. 14. Question 14 · Domain 1: Solution Design & Architecture · Single answer An associate asks whether any of a target company's 3,000 agreements contain change-of-control clauses affecting its loans. Which documents matter depends on earlier findings, and a partner reviews every conclusion. Which pattern fits best? Select the best answer. - A. A fixed workflow running the same search sequence on every review. - B. A single prompt that holds all 3,000 agreements at once in a long context window. - C. A bounded agent limited to read-only search tools. - D. An agent with write access to every document and no turn limit. 15. Question 15 · Domain 2: Claude Models, Prompting & Context Engineering · Single answer A troubleshooting agent runs up to 40 turns, and each diagnostic tool returns about 8,000 tokens of router logs that stay in the conversation. After turn 20 it forgets early constraints. What should change? Select the best answer. - A. Switch to a model with more context and leave every log in the conversation. - B. Trim used logs to short extracts; keep constraints in the system prompt. - C. Restart the session every ten turns with a summary the engineer writes by hand. - D. Ask the tools to return logs as JSON instead of text. 16. Question 16 · Domain 3: Integration · Single answer A claims agent has 27 tools, including four overlapping search tools and run_query, which runs any SQL. Wrong-tool selection is 11 percent, and security has flagged run_query. What should you do? Select the best answer. - A. Add examples of each tool's use to the prompt, with a decision table. - B. Keep run_query and make the agent explain each query first. - C. Merge the overlapping search tools and swap run_query for narrow read tools. - D. Add four more specialized tools for the gaps users report. 17. Question 17 · Domain 1: Solution Design & Architecture · Single answer A monthly safety report must synthesize 2,400 incident reports. Putting them all in one prompt overflowed the context window, and a subset produced trends nobody could trace to a report. Which approach fits best? Select the best answer. - A. Take a random sample of 300 reports each month and synthesize from it. - B. Send every report in one call, with a long-context model. - C. Summarize report groups with cited IDs before combining them. - D. Ask the model to list its sources at the end. 18. Question 18 · Domain 3: Integration · Single answer Engineers search 90,000 documents indexed with dense embeddings. Most queries contain an exact part number or fault code, and for those the top result often covers a similar but different code. What should change? Select the best answer. - A. Re-embed every document using a larger, higher-dimensional embedding model. - B. Have the model confirm the retrieved part number matches the one asked about. - C. Halve the chunk size across the whole index. - D. Add keyword search for identifiers, merged with embedding results. 19. Question 19 · Domain 2: Claude Models, Prompting & Context Engineering · Single answer A planning assistant sends an 11,000-token system prompt on every request, with caching enabled. The prompt starts with the current time and the applicant's reference, and the hit rate is 3 percent. What fixes it? Select the best answer. - A. Cache the user turn, not the system prompt. - B. Move the time and reference after the stable planning rules. - C. Lengthen the cache lifetime to one hour. - D. Add a second breakpoint and keep the time first. 20. Question 20 · Domain 1: Solution Design & Architecture · Single answer Five chained agents write product descriptions: tone, keywords, banned phrases, translation and review, then editors approve each. Costs are nine times one call, and p95 is 40 seconds against a six-second budget. What should change? Select the best answer. - A. Run the five agents in parallel and merge their output with a sixth agent. - B. Put all five agents on the fastest tier and keep the sequence. - C. Cache between agents and raise the budget to 40 seconds. - D. One call for the writing, code for the banned phrases. 21. Question 21 · Domain 3: Integration · Single answer A public assistant lets anyone report a fault. Its tools file work orders through a privileged service account and read the staff roster to name the responsible crew. What is the best change? Select the best answer. - A. Remove the roster tool and narrow the work-order credential. - B. Keep both tools and tell the assistant never to reveal staff names. - C. Require each reporter to create an account first. - D. Log each work order and roster lookup for weekly review. 22. Question 22 · Domain 2: Claude Models, Prompting & Context Engineering · Single answer On 900 rebooking cases, the fast, middle and top tiers score 78, 91 and 93 percent. The floor is 90 percent, and only the top tier exceeds the eight-second p95 budget. Which tier fits? Select the best answer. - A. The middle tier, with any later change re-tested on both limits. - B. The top tier, with the p95 budget raised to ten seconds for this assistant. - C. The fast tier, retrying on the top tier when an answer looks wrong. - D. The fast tier, the cheapest of the three. 23. Question 23 · Domain 1: Solution Design & Architecture · Single answer An operations chief sponsored a mortgage-document system to lower cost per file while keeping a five-day turnaround. Six weeks in, cost per file is down 40 percent. How should the architect present results? Select the best answer. - A. Lead with the 96 percent extraction accuracy and the 1.8-second model latency. - B. Walk through the architecture diagram and prompt design first. - C. Report cost per file and turnaround first. - D. Show a live demo of the system processing a mortgage file end to end. 24. Question 24 · Domain 3: Integration · Single answer Seven teams build AI tools that need the same student-record lookups. Each wrote its own API wrapper, three handle the consent flag differently, and the registrar cannot see who accesses records. Which integration fits? Select the best answer. - A. A written standard for the consent flag, with a shared checklist. - B. Agent-to-agent links between each team's assistants. - C. Direct database reads and writes, skipping the REST API. - D. One registrar-owned MCP server that enforces consent and logs access. 25. Question 25 · Domain 2: Claude Models, Prompting & Context Engineering · Single answer Your team sorts guest reviews into twelve house-specific labels. A zero-shot prompt with definitions scores 74 percent, and most errors fall between neighboring labels. You have 300 labeled reviews. What should you try first? Select the best answer. - A. Ask the model to reason step by step before it picks one of the labels. - B. Add few-shot examples of the confused neighboring labels. - C. Merge the twelve labels into four broader themes. - D. Set the temperature to zero for steadier labels. 26. Question 26 · Domain 1: Solution Design & Architecture · Select 2 Claude reads adverse-event emails and writes records straight into a safety database that regulators inspect. In a pilot, 7 percent of records had a wrong or missing field. Which two design changes fit best? (Select two.) Select exactly 2.**0/2 selected.** - A. Return records through a typed schema and validate them in code. - B. Ask Claude to double-check each record before writing it. - C. Move extraction to the most capable model, keeping direct writes. - D. Send records that fail validation to a reviewer with the source email. - E. Keep direct writes and store each email for audits. 27. Question 27 · Domain 3: Integration · Single answer A medical-information assistant's unhelpful-answer rate rose from 4 to 13 percent in six weeks. Errors, latency and tokens are flat, the model is unchanged, and the corpus is refreshed weekly. What monitoring is missing? Select the best answer. - A. A fixed question set scored after every corpus refresh. - B. Tighter alert thresholds on error rate and on p95 latency for the assistant. - C. More reminders for users to flag unhelpful answers in the app. - D. Per-request token counts with an alert on any big change. 28. Question 28 · Domain 3: Integration · Select 2 You are designing two integrations: a nightly job that classifies 200,000 delivery notes with no user waiting, and a partner's scheduling agent negotiating slots with yours across company boundaries. Which two choices fit? (Select two.) Select exactly 2.**0/2 selected.** - A. Run the nightly job through an interactive client with MCP tools. - B. Run the nightly job as direct API calls with batch processing. - C. Expose your internal tools to the partner's agent through an MCP server. - D. Use an agent-to-agent protocol for the partner negotiation. - E. Have both companies share one agent with access to both systems. 29. Question 29 · Domain 2: Claude Models, Prompting & Context Engineering · Single answer Fourteen assistants each copy the same 3,000-token block of style rules and disclaimers. One disclaimer change took three weeks, and two teams shipped the old text. What should you do? Select the best answer. - A. Send every team a checklist each time the disclaimer wording changes. - B. Put the block in each assistant's first user message instead. - C. Move the block into one versioned shared prompt module. - D. Shorten the block so copies are easier to fix. 30. Question 30 · Domain 1: Solution Design & Architecture · Single answer A lender's board wants generative AI. Each application takes 2.5 hours of re-keying figures from scanned documents and 20 minutes writing a memo. A regulated model makes the credit decision. Where should Claude help? Select the best answer. - A. Replacing the regulated scoring model with a Claude credit decision. - B. Extracting figures from the scans and drafting the memo for the officer. - C. An applicant chat assistant for lending policy. - D. Reviewing the scoring model's decisions and overruling weak ones. 31. Question 31 · Domain 3: Integration · Single answer An operations agent loads all 58 tool definitions, about 42,000 tokens, on every request, though a task uses three to five. Tool-selection errors are 12 percent. Which context strategy fits best? Select the best answer. - A. Keep all 58 definitions loaded but compress each description to one line. - B. Load a short tool catalog, then fetch full definitions as needed. - C. Move to a model that has a longer context limit. - D. Turn on prompt caching for the 42,000 tokens of definitions. 32. Question 32 · Domain 4: Evaluation, Testing & Optimization · Single answer A contract promises brokers a response within 4 seconds. The dashboard reports a 2.3-second mean and marks the SLA met, yet one broker logged 40 slow-response complaints. What should be measured? Select the best answer. - A. The median latency, which discounts a few slow outliers. - B. The mean latency for each broker. - C. Per-request p95 and p99 latency against the 4-second bound. - D. Monthly complaint counts and their trend by quarter. 33. Question 33 · Domain 4: Evaluation, Testing & Optimization · Single answer An admissions assistant passed 40 tests written by the developer of its prompt, yet 18 percent of sampled live answers were wrong. The registrar can spare four hours. What should you build? Select the best answer. - A. Another 200 tests written by the developer, covering fee waivers and deferrals. - B. The registrar's ratings of the 40 existing tests. - C. Model grading of live answers with no human labels. - D. A live-sampled set labeled by the registrar, with a checked model grader. 34. Question 34 · Domain 4: Evaluation, Testing & Optimization · Single answer Two days into a two-week test, a prompt variant resolves 71 percent of support chats and control 68 percent, about 300 each. Traffic peaks on Mondays; the test began Wednesday. What should you do? Select the best answer. - A. Ship the variant today and keep monitoring the resolution rate. - B. Let the test run its planned two weeks. - C. Stop the test and send next Monday's traffic to the variant. - D. Keep the test running but switch the metric to cost per chat. 35. Question 35 · Domain 4: Evaluation, Testing & Optimization · Single answer Clinicians sign off every summary draft, and some drafts now name medications absent from the notes. That week a smaller tier and wider retrieval (6 to 20 chunks) shipped together, untested. What should come first? Select the best answer. - A. Revert to the larger tier and keep 20 chunks. - B. Test each change alone on a fixed set of notes to find the cause. - C. Add a prompt line telling the model to name only listed drugs. - D. Go back to 6 chunks and keep the smaller tier. 36. Question 36 · Domain 4: Evaluation, Testing & Optimization · Single answer A tariff-code classifier scores 94 percent on its evaluation set, but a broker says a third of codes for Portuguese and Spanish descriptions are wrong. Every evaluation item is in English. What does this show? Select the best answer. - A. The model cannot handle Portuguese or Spanish input without translation first. - B. The broker's few examples cannot outweigh a 94 percent score. - C. The evaluation set never sampled the failing traffic. - D. The prompt needs examples of tariff codes. 37. Question 37 · Domain 4: Evaluation, Testing & Optimization · Single answer A weekly job regenerates 70,000 product descriptions with synchronous calls, each carrying the same 6,000-token style guide. Nobody reads the output until Monday. Finance wants cost cut by a third. What fits best? Select the best answer. - A. Keep synchronous calls and drop the style guide. - B. Send the job through the Message Batches API. - C. Move the whole job to the top model tier. - D. Run all the requests in parallel. 38. Question 38 · Domain 4: Evaluation, Testing & Optimization · Single answer A chat assistant's p95 is 9 seconds against a 4-second target. Tracing shows generation steady near 2 seconds, while an identity lookup tool sometimes takes 7. What should you do? Select the best answer. - A. Move the assistant to the fastest model tier to cut generation time. - B. Stream the reply and show the first words during the lookup. - C. Raise the p95 target to 9 seconds to match what the system does today. - D. Put a timeout, fallback and session cache on the lookup. 39. Question 39 · Domain 4: Evaluation, Testing & Optimization · Single answer A manual-search assistant logs only HTTP status and total latency. When answers grew vaguer over several weeks, nobody could tell whether the prompt, model or indexed manuals had changed. What logging fits best? Select the best answer. - A. Keep the full text of every prompt and response for a whole year. - B. Log the prompt version, model ID and chunk IDs for each request. - C. Add alerts on HTTP error rate and total latency, with tighter thresholds. - D. Add a thumbs-down button to the answer screen. 40. Question 40 · Domain 4: Evaluation, Testing & Optimization · Select 2 A law firm's assistant extracts clauses from client contracts and reports only 96 percent extraction accuracy. A client asks how it is tested for safety and security before each release. Which two metrics fit? (Select two.) Select exactly 2.**0/2 selected.** - A. How often hidden instructions in an uploaded contract change its behavior. - B. Extraction accuracy on a larger labeled set of contracts. - C. How often one client's text shows up in another client's session. - D. Average response latency at peak hours. - E. The share of lawyers who rate the assistant useful. 41. Question 41 · Domain 4: Evaluation, Testing & Optimization · Single answer You must compare 500-token and 1,200-token chunking for a fare-rules assistant. You have 600 labeled questions, and no customer may knowingly get a worse configuration during peak season. How should you compare them? Select the best answer. - A. Test both offline on the 600 questions before any live canary. - B. Split live traffic evenly for the whole peak season. - C. Pick 1,200-token chunks and skip the comparison. - D. Ask five staff which answers they like. 42. Question 42 · Domain 5: Governance, Safety & Risk Management · Single answer A member assistant can waive fees up to $50 through a tool. Pasted text told it to waive a fee on an ineligible account, and it did. Which control fits best? Select the best answer. - A. Enforce the waiver rules inside the tool's handler. - B. Add a stronger classifier on the assistant's replies to catch waiver confirmations. - C. Add a system prompt rule never to follow pasted instructions. - D. Ask the member to confirm the waiver in chat before the tool runs. 43. Question 43 · Domain 5: Governance, Safety & Risk Management · Single answer Policy forbids sending a resident's national ID number to the model provider. Residents type it in about 15 percent of chats, always in the same nine-character format. Which control fits best? Select the best answer. - A. Screen each message using a model-based classifier before forwarding it. - B. Ask residents in the chat not to type the number. - C. Strip the number from stored logs each night. - D. Redact the number with a pattern match before any model call. 44. Question 44 · Domain 5: Governance, Safety & Risk Management · Single answer An agent downloads PDFs from unknown suppliers, extracts pricing, drafts a comparison table and can email it to committee members. The risk register must be signed off first. Which risk belongs at the top? Select the best answer. - A. Committee members disagreeing with the comparison table's layout. - B. Token costs rising sharply in months with many more proposals. - C. Hidden text in a PDF steering the agent's figures or emails. - D. Downloads timing out on slow supplier websites. 45. Question 45 · Domain 5: Governance, Safety & Risk Management · Single answer A claims assistant recommends approve, request evidence or deny on 3,000 claims weekly. A wrong denial leaves a customer without a car for weeks; a wrong small approval costs little. Where should review sit? Select the best answer. - A. On a random 10 percent of all recommendations. - B. Only below 0.8 model confidence. - C. After payment, on claims customers later dispute. - D. Before any denial or large payout is issued to a customer. 46. Question 46 · Domain 5: Governance, Safety & Risk Management · Single answer A pharmacy team approves drafted medication-change letters in a median of 9 seconds; 98 percent pass unchanged. The screen shows only the draft and two buttons. Two wrong-dose letters got through. What should change? Select the best answer. - A. Require each pharmacist to spend two full minutes on every draft, with a timer. - B. Show the prescribing record beside each draft with doubts flagged. - C. Remove the review step and let model confidence decide what gets sent. - D. Add a dose reminder above the "Approve" button. 47. Question 47 · Domain 5: Governance, Safety & Risk Management · Single answer A prescription assistant runs under a business associate agreement. Before a HIPAA audit, you find its observability platform, whose vendor has signed no such agreement, stores full prompts with patient data. Which conclusion is right? Select the best answer. - A. In scope: redact the logs or move them to a covered store. - B. Only the model provider is in scope; the logs can stay as they are. - C. The logs are fine provided the observability platform requires single sign-on. - D. The provider's agreement covers every vendor in the path, logs included. 48. Question 48 · Domain 5: Governance, Safety & Risk Management · Select 2 A student-support assistant is extending to EU applicants. Each request sends the full applicant record, including disability and nationality, and every transcript is kept for evaluation. Which two controls address GDPR most directly? (Select two.) Select exactly 2.**0/2 selected.** - A. Encrypt the full record in transit and keep sending all of it. - B. Send only the fields each question needs. - C. Keep every transcript indefinitely and anonymize it only if a regulator asks. - D. Add a privacy notice to the chat and keep the current data flow. - E. Record the lawful basis and make erasure cover stored transcripts. 49. Question 49 · Domain 5: Governance, Safety & Risk Management · Single answer A loan pre-screening assistant drafts recommendations that officers confirm. Protected attributes are absent, but ZIP code and employer are present. A risk committee asks what shows the assistant treats groups fairly. What should you provide? Select the best answer. - A. Outcomes per applicant group measured against a defined fairness metric. - B. A statement that protected attributes never appear in any prompt the assistant sees. - C. Overall approval rates matching last year's baseline. - D. The model vendor's own published bias evaluation report. 50. Question 50 · Domain 5: Governance, Safety & Risk Management · Single answer After three months, maintenance engineers accept 97 percent of drafted fault diagnoses without opening the logs, and two incidents trace to wrong diagnoses accepted unchanged. Which failure mode is this, and what mitigates it? Select the best answer. - A. Model drift; move to a newer model and compare. - B. Hallucination; lower the temperature. - C. Alert fatigue; send engineers fewer drafts per shift. - D. Automation bias; show the log evidence behind each draft. 51. Question 51 · Domain 6: Stakeholder Communication & Lifecycle Management · Single answer An operations chief wants an assistant to answer all 2,500 daily customer emails and asks for a proposal in two weeks. The emails span bookings, disputes, invoices and damage claims. What should discovery establish first? Select the best answer. - A. Which model tier and which prompt style would suit the assistant best. - B. How quickly a working prototype could answer every type of email. - C. Each email type's volume and the cost of a wrong reply. - D. Whether customers would rather use chat than email, and for which requests. 52. Question 52 · Domain 6: Stakeholder Communication & Lifecycle Management · Single answer A finance chief must choose between calling the Claude API directly and using it through a cloud platform the company already uses. The decision memo lists only each route's benefits. What should it add? Select the best answer. - A. What each route gives up and what reversing it later would cost. - B. A per-token price comparison of input and output costs on both routes. - C. A longer list of each route's benefits, gathered from both vendors' websites. - D. One recommendation with the alternatives left out. 53. Question 53 · Domain 6: Stakeholder Communication & Lifecycle Management · Single answer An editor proposes that 99 percent of suggested headlines be "acceptable" and that suggestions return within 2 seconds. The pilot measured a lower acceptance rate, and nobody has defined "acceptable". What should the architect do? Select the best answer. - A. Accept both terms as written and improve the model in the weeks before launch. - B. Drop the latency term and keep the 99 percent target. - C. Cut both targets by ten percent to leave a margin. - D. Tie each threshold to pilot evidence and define how "acceptable" is scored. 54. Question 54 · Domain 6: Stakeholder Communication & Lifecycle Management · Single answer Dashboards show latency, errors, cost and a weekly evaluation score. Over six weeks the score fell from 88 to 79 with errors flat, and nobody acted until a plant manager complained. What governance is missing? Select the best answer. - A. A named owner who must act when each signal crosses its threshold. - B. A new dashboard that combines all four signals on one screen. - C. Daily dashboard emails to every plant manager. - D. A more sensitive evaluation that catches smaller drops. 55. Question 55 · Domain 6: Stakeholder Communication & Lifecycle Management · Single answer An inherited assistant came with a diagram, an API reference and a runbook. A new engineer raised the chunk size and unknowingly broke a control that kept account data out of retrieval. What was missing? Select the best answer. - A. A more detailed architecture diagram showing every component and data flow. - B. A recorded video walkthrough of the assistant's user interface. - C. A longer runbook covering more incident types. - D. A decision log of why each setting exists and which controls depend on it. 56. Question 56 · Domain 6: Stakeholder Communication & Lifecycle Management · Single answer Two months after launch, the partner has left, the model faces deprecation, the catalog has changed and the evaluation score fell from 90 to 84. Nobody owns the assistant. What is missing? Select the best answer. - A. A longer contract with the partner to keep adding new features after the launch. - B. A much larger pre-launch test set, run again before launch. - C. Automatic switching to each new model version on release. - D. An owned monitoring phase with triggers for redesign. 57. Question 57 · Domain 6: Stakeholder Communication & Lifecycle Management · Select 2 A city permit office takes 11 weeks to decide building permits, but state law requires 8. It wants design to start now, and you insist on discovery first. Which two questions must discovery answer? (Select two.) Select exactly 2.**0/2 selected.** - A. Which chat channel and update format applicants prefer. - B. Which model tier the office's budget will support. - C. Which decisions the law reserves for a permit official. - D. How many permits the office could decide with no staff at all. - E. Which stages and causes account for the 11 weeks. 58. Question 58 · Domain 6: Stakeholder Communication & Lifecycle Management · Single answer You chose a fixed workflow with one Claude summarization step for fault triage over the autonomous agent a vendor demonstrated. A non-technical board asks you to justify it. What should you lead with? Select the best answer. - A. Benchmark scores showing the workflow beats the agent on a public test. - B. A live demo of the workflow handling a real fault, like the vendor's. - C. The business constraint behind the choice and its trade-off. - D. The model tier and prompt design you chose. 59. Question 59 · Domain 6: Stakeholder Communication & Lifecycle Management · Single answer A sponsor says a lease-summary assistant must be "always right" before launch. The pilot shows 93 percent field accuracy, and most errors fall in one clause type the review step catches. What should you do? Select the best answer. - A. Delay launch until it scores 100 percent. - B. Propose a measured accuracy threshold backed by the pilot evidence. - C. Agree to "always right" and quietly rely on the review step. - D. Drop that clause type from scope without telling the sponsor. 60. Question 60 · Domain 7: Developer Productivity & Operational Enablement · Single answer Forty engineers use Claude Code across six repositories with personal instruction files, and some allow any shell command. The platform lead wants shared conventions and safe permissions. What should you set up? Select the best answer. - A. A shared wiki page of conventions that each engineer copies into a personal file. - B. A rule that every command needs manual approval, with nothing saved. - C. A committed project CLAUDE.md and settings file per repository. - D. One personal file shared and emailed to the team. 61. Question 61 · Domain 7: Developer Productivity & Operational Enablement · Single answer Pull requests doubled after a signaling team adopted AI-assisted coding, and two regressions reached production from approved AI-generated changes. Reviewers could not explain either change. The safety manager wants a decision. What fits best? Select the best answer. - A. Keep the tool but make authors explain every change they submit. - B. Withdraw the tool from the whole signaling codebase and go back to manual coding. - C. Leave the current review process unchanged for now. - D. Require a second reviewer on every AI-generated change. 62. Question 62 · Domain 7: Developer Productivity & Operational Enablement · Single answer About 3 percent of an assistant's calls to the core banking system fail, mostly when branches open. Each failure returns HTTP 429 within 200 milliseconds. What should you do? Select the best answer. - A. Report an outage to the core system team and ask for more servers. - B. Back off retries under a concurrency cap. - C. Raise the tool's timeout from 5 seconds to 30 seconds. - D. Switch the assistant to a faster model tier for the morning peak. 63. Question 63 · Domain 7: Developer Productivity & Operational Enablement · Single answer A team of 25 engineers will adopt Claude Code this quarter. Several repositories hold system credentials in environment files, and the CIO wants measurable value without a security incident. What should come first? Select the best answer. - A. Give all 25 engineers full access on day one and track adoption. - B. Run company-wide training, then let engineers install the tool. - C. Set team permission and sandbox rules before any repository connects. - D. Let each engineer pick their own settings. Before you ask ## Before you sit the mock **Is this the real CCAR-P exam?** No. Every item here was written for this page from the official CCAR-P Exam Guide and its published domains and objectives. Nothing is recalled from a sitting, nothing is drawn from another provider, and Cred Farmer is not affiliated with Anthropic or Pearson VUE. The items are short scenarios, about the length of the guide's sample questions. Of the 63, 57 are single-answer with four options and 6 ask you to select two of five, saying so in the stem. The questions themselves are ours, and their difficulty is not calibrated against the exam. **How is the mock scored?** You get the raw count of correct answers out of the items you sat, and the same count for each domain beside that domain's official weight. A select-two item counts only when both picks match the key. The exam reports a scaled score with a 720 cut, and no practice percentage converts to it, so this page never shows a scaled score, a pass verdict or a percentile. **Do I need an account to take it?** No. Choose a start mode and the sitting runs in your browser tab. There is no sign-up, no email field and no cookie that follows you. An account only matters if you want the larger signed-in library, and the link for that sits at the bottom of the page rather than in your way. **Is it actually free, or is there a catch?** No catch. Everything on Cred Farmer is free, this page included: no card, no trial, no limit on how many times you sit it. An account costs nothing either; it adds the five timed forms, saved progress and a readiness score capped by your first sitting of each form. The one limit here is inventory: the no-account paper is one paper, so sit it a second time and you will remember some items, and that is when the signed-in forms become the next step. **How many questions are in the mock?** 63 in the full sitting, allocated across the 7 official domains in proportion to their weights. The clock runs at the official pace of 63 items in 120 minutes, about 1.9 minutes an item, so the full sitting gets 120 minutes, the same as the exam. The quick sitting draws 30 of them evenly across the domains in 57 minutes, and the untimed review takes the whole paper without a clock. **What happens to my result?** It stays in this tab. Nothing is sent to a server, nothing is stored on your device, and nothing you answer here feeds a readiness score or any aggregate. Close or refresh the page and the result is gone, which is also why it is worth reading the per-option reasons before you leave. ## This page is 63 items. The account has 315. Everything on Cred Farmer is free. An account adds the five timed forms, saved progress and a readiness score: CCAR-P has 5 forms aligned to the current blueprint, 63 items each, 315 in total, plus drills by objective, and your first sitting of each form sets the band the score can reach. No card, no trial clock. [Create a free account for the five timed forms](https://credfarmer.com/start/ccar-p?to=practice) Every item on this page was written for it from the official CCAR-P Exam Guide, not recalled from a sitting and not copied from anyone else's set. Material sold as recalled questions can cost you the credential: read [why Claude certification exam dumps are a bad bet](https://credfarmer.com/blog/claude-certification-exam-dumps). Items were last checked against the guide on 17 September 2026. The guide is published on Anthropic's [CCAR-P certification page](https://anthropic-partners.skilljar.com/claude-certified-architect-professional-certification). Not affiliated with, or endorsed by, Anthropic or Pearson VUE. Exam facts are summarised from publicly published program information and can change. Always confirm against the official exam guide before booking.