Practice questions
110 scenario questions across all seven domains, mixed by default so you have to recognise which domain a question belongs to — half the difficulty on the real exam. Every answer explains why the wrong options are wrong, which is where the reasoning actually lives.
0 / 0correct
0current streak
0best streak
Browse all 110 questions
Each question has its own page with the answer and a full explanation.
Integration 21
- A marketing-analytics agent is connected to a vendor's MCP server that exposes 30 tools, including campaign…
- A coding agent used in CI only needs to run the test suite and read repository files. It currently runs with…
- A support organization's agent needs to issue goodwill credits in about 5% of conversations; most…
- A legal-research assistant serves attorneys at a firm with strict ethical walls: attorneys on one matter must…
- A banking assistant's get_transactions tool accepts an account_id argument filled in by the model from the…
- A platform team moves an internal MCP server that lets agents query HR and payroll data from a local stdio…
- A fraud-screening step inside a checkout flow must return an allow, deny or review decision in under 800 ms…
- A document-QA assistant answers employee questions. On your evaluation set, a fast model matches a larger…
- An insurance company's nightly job extracts fields from 200,000 scanned claim letters for next-morning…
- A multi-agent customer-onboarding system intermittently takes over a minute to respond. Each service logs its…
- Finance asks why monthly Claude spend for a multi-tenant SaaS product rose 70% while active users rose 10%…
- A healthcare-scheduling agent produces about 2 million traces per day. The observability bill is growing…
- A pharmaceutical company's regulatory-QA assistant splits submission documents into fixed 400-token windows…
- After a quarterly refresh of an internal knowledge base, an engineering-support assistant began recommending…
- A team upgrades its embedding model and, to save time, embeds only newly added documents with the new model…
- An IT-service assistant answers from a single vector index of knowledge articles and resolved tickets…
- A sales-operations assistant is asked questions like "what was our average deal size in EMEA last quarter, by…
- Fourteen product teams are each building agents that need to look up customer entitlements from the same…
- A logistics company's planning agent must work with a carrier partner on complex multi-leg shipments. The…
- A company-wide assistant has 40 Skills installed, from expense policy to incident response. Users report it…
- A compliance-review agent has a short, stable set of non-negotiable rules (about one page) that must apply to…
Solution Design & Architecture 19
- A hospital network's COO asks for "a Claude agent to fix discharge delays." Discovery shows physicians spend…
- An insurer's eligibility rules are written as explicit thresholds in an underwriting manual, but applications…
- Purchase orders take nine days end to end. Analysis shows about seven of those days are spent waiting for…
- A retail bank's contact-center business case lists lower cost, faster answers and "transformation" as goals…
- A legal team's pilot report claims a 70% time saving for a contract-review assistant, comparing the time…
- A marketplace search assistant must meet a contractual p95 latency limit with a partner. The team proposes a…
- An invoice-extraction pipeline tested well, but in production about 15% of outputs are empty or nonsensical…
- Claude drafts refund decisions that a downstream system executes. Policy says refunds above a threshold need…
- A support-answer assistant collects thumbs-up/down ratings from agents. A dashboard shows satisfaction…
- HR wants employees to ask questions about roughly 200 benefits policy documents. Requirements are grounded…
- Select two tasks where an agentic pattern is better justified than a predefined workflow.
- A telecom support agent handles every contact. About 80% of contacts are six well-known intents; the rest are…
- A team split contract drafting into six agents, one per clause type. Clauses cross-reference each other and…
- A research coordinator fans out to eight search workers, each given the full topic. Final reports contain…
- A customer-operations system has an intake classifier, a research worker, a resolution drafter and an…
- A multi-agent analysis system produced a wrong final answer. Each agent writes its own logs with its own…
- Reviewing a 60-file migration pull request in one pass missed issues in the middle files. The team then tried…
- Claude produces quarterly variance commentary by computing variances from a ledger export and then explaining…
- An architect must assess an unfamiliar legacy codebase for cloud-migration readiness. A consultant has…
Evaluation, Testing & Optimization 18
- A bank is preparing to launch a Claude-based assistant that answers customer questions about account fees…
- An engineering lead reports that a new configuration of a document-processing agent reduced cost per API call…
- A healthcare provider's patient-intake assistant passed launch review with a zero rate of disallowed medical…
- A team building a contract-review assistant has an evaluation set of 200 questions written by its prompt…
- A customer-support assistant must (1) return a JSON object matching a fixed schema, (2) call the order-lookup…
- An LLM judge is used to compare two versions of a research summarizer pairwise. The new version wins 68% of…
- An agent that files insurance claims passes a test case on its first run, and the team marks the case green…
- A travel company's booking assistant is a multi-turn chat. The team wants to A/B test a new system prompt and…
- After an online test, a new retrieval configuration for an internal IT helpdesk assistant improves the…
- A team is comparing two prompt variants for a claims-summary feature over a four-week online test. Both…
- A pharmaceutical company's RAG assistant answers questions about drug-interaction guidance. After the…
- A procurement agent's responses intermittently end mid-sentence when it summarizes long supplier contracts…
- A logistics company routes all customer email through a large, capable model that classifies each message…
- An internal policy assistant started giving inconsistent answers about expense limits a week ago: the same…
- A media company runs two Claude workloads on the same large model: a real-time newsroom assistant, and a…
- A customer-service chat assistant has a p95 latency complaint from users, who mostly perceive the wait before…
- A financial-services firm's research assistant shows healthy dashboards: 99.9% uptime, low HTTP error rates…
- A SaaS company's monitoring shows that, after a release, cost per conversation rose 30% while traffic, model…
Governance, Safety & Risk Management 15
- A wealth-management firm deploys a Claude agent that answers advisors' questions about client portfolios. Its…
- An agent triages inbound customer emails, summarizes them, and has tools to look up orders, issue store…
- A platform team runs a Claude-based internal agent for IT operations. It has tools for reading tickets…
- An insurer's policy Q&A assistant has run well for months. The week after the content team migrated the…
- You are building a risk register for an agent that handles accounts-payable exceptions. Four candidate risks…
- A multi-step research agent compiles supplier risk reports. Reviewers notice that when one data-source tool…
- A healthcare payer's prior-authorization assistant proposes approve or deny recommendations. Clinical…
- A logistics company's agent can reschedule deliveries, issue refunds up to a configured limit, and change a…
- A team implemented an approval step for high-value contract amendments: the agent drafts the amendment and a…
- A European retailer's support assistant stores conversation transcripts, writes full prompts to application…
- A US clinic network's legal team has confirmed a business associate agreement covering the Claude model…
- A federal agency contractor is designing a Claude-based document analysis system for data categorized at the…
- A lender pilots Claude to produce recommended credit-limit decisions for small-business applicants in the EU…
- A recruiting platform uses Claude to rank applicants against job descriptions. To address bias concerns, the…
- A public-sector benefits agency uses Claude to draft eligibility determinations that caseworkers finalize…
Stakeholder Communication & Lifecycle Management 15
- A retail bank's VP of operations asks you for 'an agent that handles card disputes end to end' and wants a…
- A feasibility probe for invoice extraction scored very well on 30 invoices the finance team picked as…
- You are running discovery for an assistant that will answer employees' leave and benefits questions in both…
- A claims-triage system routes drafts to adjusters. Product wants to stop human review for low-value claims to…
- You are briefing an executive committee on a drafting assistant for customer letters before a go-live…
- Eight months after launch, a new model version is released and the team debates whether the original choice…
- A business unit wants you to commit to an availability target for a customer-facing assistant that is…
- Which of the following is the best-defined quality objective for an HR policy assistant?
- An executive forwards a screenshot of one clearly wrong answer from the internal knowledge assistant and asks…
- A claims assistant answers most customer queries in seconds, but customers complain that some responses take…
- Several product teams will build Claude features on your platform architecture. In earlier projects, teams…
- A build team is about to hand a Claude-based underwriting assistant to the platform operations team. Which…
- Overnight, the daily cost of a multi-step Claude pipeline rises sharply with no change in request volume. You…
- After launch, an engineer edited the production prompt to make a document assistant's tone warmer. A week…
- Your production system is pinned to a model version, and the provider announces that version will be retired…
Claude Models, Prompting & Context Engineering 14
- A personal-finance app categorizes every card transaction with Claude, millions of events a month, and the…
- A contract-analysis assistant on the mid tier keeps giving wrong termination dates. Tracing shows retrieval…
- You are designing a cascade for an internal IT help assistant: the smallest tier answers first, and some…
- A document-extraction service calls Claude through a model alias that always resolves to the latest version…
- An employee self-service assistant answers pay questions. Its HR database tool runs under a service account…
- A SaaS vendor runs a Claude assistant for 30 enterprise tenants. Each tenant has a hand-edited copy of the…
- A vendor-research agent fetches supplier web pages, reads internal procurement files for context, and emails…
- A marketplace uses Claude to assign a condition label (new, refurbished, used, for parts) to listings. The…
- A benefits-eligibility step applies six interacting conditions, and errors cluster on cases where several…
- A lender uses Claude, with a reasoning section before each answer, to draft credit-limit recommendations that…
- A team enabled prompt caching with a cache_control breakpoint after the system prompt. Usage data shows cache…
- A support agent handles long troubleshooting conversations. To control tokens, older turns are replaced with…
- An enterprise assistant's system prompt carries 25 departmental procedures (expense filing, visa letters…
- A nightly account-research agent's cost has doubled. A token breakdown shows most input tokens are raw JSON…
Developer Productivity & Operational Enablement 8
- Security requires that Claude Code never read files under a repository's secrets/ directory. The rule is…
- New hires report that Claude Code ignores the team's test command and naming conventions, while the tech lead…
- A platform team wants every developer to use the same issue-tracker and read-only analytics-database MCP…
- Developers complain about constant permission prompts in Claude Code, and several have added a rule allowing…
- After adopting Claude Code, a team's merged pull requests per week rose sharply, but lead time to production…
- A team runs Claude Code non-interactively in CI to triage issues filed on its public repository and propose…
- An intermittent date-handling bug keeps returning. Claude Code has proposed three fixes in succession; each…
- An SRE organization is designing a Claude-based incident-triage assistant. Which design is most appropriate?