Skip to Content
View all events

Claude Certified Architect – Foundations Exam Prep: A Two-Hour Crash Course

Published by O'Reilly Media, Inc.

Intermediate content levelIntermediate

Decode the five exam domains, drill the six official production scenarios with sample questions, and leave with a targeted study plan

What you’ll learn and how you can apply it

  • Map the CCAR-F exam blueprint
  • Design agentic architectures the exam expects
  • Apply MCP and tool-design judgment
  • Configure Claude Code the way the scenarios demand
  • Enforce structured output and reliability patterns
  • Build a personal study plan

Course description

In this two-hour course, you’ll prepare for Anthropic's Claude Certified Architect – Foundations (CCAR-F) exam. You’ll start by decoding the exam blueprint—the five weighted domains and the six official production scenarios that frame every question. Then you’ll work through each domain the way the exam tests it, drilling official-style sample questions, and dissecting why each distractor is wrong. By the end, you’ll have a domain-by-domain study checklist, hands-on familiarity with the judgment calls the exam rewards, and a concrete plan for scheduling and passing CCAR-F.

This live event is for you because...

  • You’re a solution architect or AI engineer designing production Claude applications and you want the CCAR-F credential to validate your architectural judgment.
  • You’re a software engineer who uses Claude Code daily but needs to formalize your knowledge of configuration hierarchies, orchestration patterns, and the trade-offs the exam tests.
  • You’re a tech lead or consultant at an Anthropic partner organization whose partner tier or client engagements benefit from certified staff, and you need an efficient, structured prep path.

Prerequisites

  • Hands-on experience building with Claude (Claude API, Claude Code, or the Claude Agent SDK)
  • Familiarity with core LLM concepts
  • Comfort reading Python or TypeScript snippets
  • A laptop with Claude Code installed (helpful for following the demos but not mandatory)

Recommended preparation:

Recommended follow-up:

Schedule

The time frames are only estimates and may vary according to how the class is progressing.

Inside the CCAR-F Exam and Agentic Architecture and Orchestration (35 minutes)

  • Presentation: How the exam works; 60 multiple-choice/multiple-response items in 120 minutes; the scenario format; the five-domain blueprint and weights, what is explicitly out of scope, and the 12-month credential validity with free on-time renewal; Domain 1, Agentic Architecture and Orchestration; the agentic loop; coordinator–subagent orchestration; hooks for deterministic control; session management; when to start fresh with an injected summary instead
  • Hands-on exercise: Domain 1 question drill
  • Q&A
  • Break

Domain 2: Tool design and MCP integration (20 minutes)

  • Presentation: Tool descriptions as the primary selection mechanism; structured MCP errors; access failures versus valid empty results; tool distribution; MCP configuration
  • Demonstration: A well-described versus minimally-described tool pair in a live agent—watching misrouting happen, then fixing it with description quality alone
  • Hands-on exercise: Domain 2 question drill
  • Q&A

Domain 3: Claude Code Configuration and Workflows (25 minutes)

  • Presentation: The CLAUDE.md hierarchy; slash commands and skills; plan mode versus direct execution: the complexity rubric the exam applies and the explore subagent for verbose discovery; CI/CD integration and why an independent review instance beats self-review
  • Demonstration: Live Claude Code walkthrough
  • Hands-on exercise: Domain 3 question drill
  • Q&A
  • Break

Domains 4 and 5: Prompt Engineering, Structured Output & Context Management (30 minutes)

  • Presentation: Domain 4—Prompt Engineering and Structured Output; explicit criteria over vague instructions and managing false-positive categories to preserve developer trust; few-shot examples for ambiguous cases; structured output via tool_use with JSON schemas, tool_choice, and nullable fields to prevent fabricated values; validation-retry loops and their limits; the Message Batches API and the batch versus synchronous judgment call; multi-instance and multi-pass review; Domain 5—Context Management and Reliability; “Lost in the middle” effects, trimming verbose tool outputs, persistent case-facts blocks, scratchpad files, and /compact for long sessions; escalation design— explicit triggers (customer demands a human, policy gaps, no progress) versus unreliable proxies like sentiment or self-reported confidence; structured error propagation across multi-agent systems and human review workflows with stratified sampling and calibrated field-level confidence
  • Hands-on exercise: Mixed-domain question drill
  • Q&A

Your study plan and exam-day logistics (10 minutes)

  • Presentation: A domain-by-domain study checklist that maps the exam guide’s four official prep exercises to your weakest domains from today’s drills; registration walkthrough—Anthropic Partner Academy → Pearson VUE, the 24-hour reschedule rule, retake waiting periods, the NDA, and the free 12-month renewal assessment
  • Q&A

Your Instructor

Lucas Soares

Lucas Soares is a machine learning engineer who has worked at K1 Digital and Biometrid, where he developed computer vision and NLP models for applications such as document verification, OCR-based applications, and recommender systems. Lucas has also developed various ML models, including neural networks, Siamese networks, convolutional neural networks, LSTMs, and genetic algorithms.

linkedinXlinksearch

Skills covered

  • Claude
  • QA / Testing