Claude Cert Arch Domain 1 — Development skill for Claude Code

Claude Certified Architect - Foundations: Agentic Architecture & Orchestration


This is the companion repo for Claude Certified Architect - Foundations: Agentic Architecture & Orchestration on Pluralsight. It covers Domain 1 of the CCAR-F exam, the largest at 27 percent: agentic loops, multi-agent coordination, subagent context passing, programmatic enforcement with hooks, task decomposition, and session resumption and forking.

Each module has demo code you'll watch being built in the course, a hands-on challenge with a broken starter to fix, a worked solution, and a self-check with exam-style questions. Everything runs in GitHub Codespaces with one secret, your `ANTHROPIC_API_KEY`.

This repo is not affiliated with or endorsed by Anthropic. The CCAR-F exam guide is Anthropic's; the code and questions here are original course material.

Run it in GitHub Codespaces (recommended)

1. Add your API key as a Codespaces secret named `ANTHROPIC_API_KEY` (GitHub → Settings → Codespaces → Secrets), or accept the prompt when the Codespace starts.

2. Click **Code → Create codespace on main**.

3. Wait for the container to build. Dependencies install automatically.

4. Open a terminal and run any demo, for example:

   cd module-01-agentic-loops/demo-support-agent-loop
   python agent.py

No virtual environment needed. The container is the isolation.

Run it locally

python -m venv .venv && source .venv/bin/activate
pip install -r requirements.txt
export ANTHROPIC_API_KEY=sk-ant-...

Python 3.10 or later. The devcontainer uses 3.12.

Module 7 has you type `claude` directly in a terminal, so you'll need the CLI on your PATH for that module:

npm install -g @anthropic-ai/claude-code

Everything else uses the Claude Agent SDK, which ships its own copy of the CLI. Module 1 doesn't use the SDK at all; it runs on the `anthropic` package alone.

Repo layout

module-01-agentic-lo