Exam overview
What the Claude Certified Architect — Foundations exam covers, how it is scored, and how to plan your preparation around the blueprint.
Format and scoring
- 60 multiple-choice questions, each with a single correct answer.
- 120 minutes — roughly two minutes per question, which is generous for recall items and tight for the longer scenarios.
- 720 out of 1000 scaled to pass, so aim for consistent mock scores comfortably above that rather than scraping the line once.
- Questions are scenario-anchored: most present a situation and ask for the best action, not a definition.
Domain weights
The weights below drive how this site's mock exam samples questions, so a practice run has the same shape as the real thing.
D1Agentic Architecture & Orchestration
27% · ~16 questionsChoosing between a single call, a workflow, and an agent; the tool-use loop; sub-agents; session and state design.
D2Tool Design & MCP Integration
18% · ~11 questionsWriting tool schemas and descriptions, scoping the tool surface, server vs client tools, and wiring MCP servers.
D3Claude Code Configuration & Workflows
20% · ~12 questionsCLAUDE.md, settings and permissions, hooks, skills, sub-agents, and putting Claude Code into CI.
D4Prompt Engineering & Structured Output
20% · ~12 questionsSystem prompts, examples, structured outputs, strict tool schemas, effort and thinking, and the Batch API.
D5Context Management & Reliability
15% · ~9 questionsPrompt caching, compaction, context editing, memory, error handling, stop reasons, and retries.
Who it is aimed at
The exam targets practitioners with meaningful hands-on production experience — people who have built agentic applications, configured Claude Code for a team, designed tool and MCP interfaces, and integrated Claude into pipelines. It rewards judgement built from having shipped something, not memorisation.
The most common failure mode is knowing the features but not the trade-offs: when an agent is the wrong tier, when a hook beats an instruction, when caching silently stops working. Those judgement calls are what the scenario questions test.
How to prepare with this site
- Read the study guide end to end once, to see the shape of the whole blueprint before drilling any part of it.
- Work through practice quizzes one domain at a time. The explanations matter more than the score — read them even when you were right, since some correct answers come from good instinct rather than a solid reason.
- Use flashcards for the exact values that decide close questions: parameter names, limits, and which beta header goes with which feature.
- Take a timed mock exam only once you are scoring well in practice. Then use the per-domain breakdown to pick your next drill.
- Skim the anti-patterns the day before. Many distractors are drawn from exactly those mistakes.
A note on sources
This is an independent study aid. Everything here is written against the public Claude platform documentation, which moves quickly — when a detail here disagrees with the official docs, the docs win. Each question links to the relevant page so you can check for yourself.