Udemy
Turn what you know into an opportunity and reach millions around the world.
Learn More
Your cart is empty.
Keep shopping
The Complete Claude Certified Architect - Foundations Course
New
Rating: 5.0 out of 5(12 ratings)
102 students

The Complete Claude Certified Architect - Foundations Course

Pass the CCA-F (CCAR-F) exam on your first attempt with agent, MCP, and Claude Code demos, labs, and 2 full mock exams
Created byBiswajit | Expanding Horizons
Last updated 10/2026
English

What you'll learn

  • Prepare to pass the Claude Certified Architect Foundations (CCA-F) exam on your first attempt by focusing on exactly what the five domains cover
  • Build agent loops with the Claude API and Agent SDK, reading stop_reason and setting tool_choice to control when Claude uses tools and when it stops
  • Design multi-agent systems where a coordinator hands work to subagents and passes context between them without losing key details
  • Enforce business rules like refund limits with hooks and code checks, and know when a prompt instruction alone is not enough
  • Set up Claude Code with CLAUDE . md files, rules, settings, slash commands, and Skills, then run it in CI/CD pipelines to automate code reviews
  • Get reliable structured output with JSON schemas, few-shot examples, and retry loops, and use the Message Batches API for jobs that can wait
  • Connect Claude to external systems with MCP servers, and write clear tool descriptions and error messages that help Claude pick the right tool
  • Manage long conversations, keep critical facts from getting lost, and hand off to a human at the right moment so your agents stay reliable
  • Apply what you learn in hands-on labs where you build a support agent, a multi-agent research pipeline, a CI review job, and a data extraction pipeline
  • Practice with domain quizzes and two full 60-question mock exams, and learn to rule out tempting wrong answers across all six official exam scenarios

Course content

7 sections • 16 lectures • 4h 17m total length

  • Hi Everyone :)8:16
  • IMPORTANT NOTE1:28
  • The Exam Blueprint21:57
  • LLM & Claude Foundations for Beginners29:50

Requirements

  • Basic Python skills, like reading functions, loops, dictionaries, and JSON. No prior experience with Claude, AI agents, or MCP is needed
  • Comfort running simple commands in a terminal, like PowerShell on Windows or Terminal on macOS and Linux
  • A Windows, macOS, or Linux computer where you can install VS Code, Python, and Claude Code. Setup is covered step by step in the course
  • A Claude API account with a small amount of credit for the labs. Demos default to the budget-friendly Haiku model to keep costs down

Description

The Claude Certified Architect Foundations exam (CCAR-F, also called CCA-F) does not test syntax. It asks you to pick the right design for production scenarios, 60 times in 120 minutes. This course trains you to make those calls and pass on your first attempt.

Every topic starts with a failure from one of the six official scenarios, like a support agent approving a refund it should have blocked. Then you see the fix in a real VS Code project, not notebooks, since that is where Claude Code, MCP, and CI run. Exam Trap breakdowns show the tempting wrong answer and why it loses.

How we cover all five exam domains

  • Agentic Architecture and Orchestration (27%): agent loops and stop_reason, coordinator and subagent designs, context passing, and hooks that enforce rules a prompt cannot guarantee.

  • Tool Design and MCP Integration (18%): tool descriptions Claude picks correctly, helpful error responses, and MCP servers with tools, resources, and prompts.

  • Claude Code Configuration and Workflows (20%): CLAUDE .md, rules, settings, slash commands, and Skills in a real demo repo, plus Claude Code in CI/CD.

  • Prompt Engineering and Structured Output (20%): a five-step ladder from clear instructions up to schema enforcement, plus validation retry loops and the Message Batches API.

  • Context Management and Reliability (15%): what context costs, why a bigger window is not always the fix, keeping key facts intact in long sessions, and when to escalate to a human.

Practice and labs

Lessons end with exam-style questions where we eliminate wrong answers out loud. You also get domain quizzes and two 60-question mock exams that mirror the real exam, with detailed explanations.

In hands-on labs, you build a support agent, a multi-agent research pipeline, a CI review job, and a data extraction pipeline. The demo code defaults to the low-cost Haiku model to keep API costs down.

The course sticks to exam scope, so fine-tuning and model training are mentioned, not taught, and it stays updated as Claude Code and the exam change.

Watch the free previews to see if my teaching style suits you.

This is an independent prep course, not affiliated with or endorsed by Anthropic.

Who this course is for:

  • Anyone preparing for the Claude Certified Architect Foundations (CCA-F) exam who wants to pass on the first attempt, even if they are new to Claude
  • Developers and engineers building with the Claude API, Claude Code, or MCP who want to make solid design decisions in real projects
  • Architects, tech leads, and consultants who design or review AI agent systems and want an Anthropic certification to prove their skills

Report abuse