Claude Certified Architect – Foundations — Claude Cert Academy

Pass the only official Claude certification.

The premier exam-prep course for the Claude Certified Architect – Foundations certification. 5 domain modules mapped to the official exam blueprint, scenario-based practice drills, and a full 60-question proctored-style capstone exam.

What you'll learn

• Design and implement multi-agent systems using the Claude Agent SDK
• Configure Claude Code workflows with CLAUDE.md hierarchy and custom skills
• Write tool schemas that eliminate model confusion and handle errors reliably
• Enforce structured output with JSON schemas and retry loops
• Pass the CCA-F exam: 60 questions, 720/1000 passing score

Course content

01. Agentic Architecture & Orchestration

• Agentic loop lifecycle & stop_reason values
• Coordinator–subagent hub-and-spoke pattern
• Subagent spawning, context passing & isolation
• Multi-step workflows & handoff enforcement
• Agent SDK lifecycle hooks & tool call interception
• Task decomposition strategies for complex workflows
• Token economics & context leakage prevention
• Scenario: Debug a code-review multi-agent pipeline
• Scenario: Build a code-review multi-agent pipeline

02. Tool Design & MCP Integration

• Writing tool descriptions the model reads correctly
• Disambiguating similar-sounding tools
• Structured error responses & the isError flag
• Scoping tools per agent role
• MCP server configuration & the connection architecture
• Retryable vs. fatal error categories
• Scenario: Fix a support agent calling the wrong tool
• Scenario: Fix a support agent calling the wrong tool

03. Claude Code Configuration & Workflows

• CLAUDE.md hierarchy — how rules compose across scopes
• Custom slash commands & agent skills
• Path-scoped rules — precision over global noise
• Plan mode vs. direct execution — when collaboration beats speed
• CI/CD pipeline integration — where Claude Code fits and where it must not
• /compact, /memory and session context management
• Scenario: CLAUDE.md scoping mismatch in a monorepo
• Scenario: Configure a monorepo Claude Code workspace

04. Prompt Engineering & Structured Output

• JSON schema enforcement — programmatic contracts over prompt hopes
• Few-shot design that generalizes beyond your examples
• Validation retry loops & targeted error feedback
• XML structuring for complex prompts
• Categorical review criteria — binary over adjective
• Programmatic vs. prompt-based enforcement — choosing the right layer
• Scenario: Schema enforcement in a data extraction pipeline
• Scenario: Enforce structured output in a data extraction pipeline

05. Context Management & Reliability

• Long-Context Pitfalls & Attention Degradation
• Handoff Patterns Between Agents
• Confidence Calibration & Asking for Clarification
• The Safety Model: Harms, Refusals, and Mitigations
• Memory Architectures: In-Context, External, and Hybrid
• Scenario: The Fading Agent — Context & Reliability in a Long-Running Pipeline
• Scenario: Debug a context leak in a long-running agent

Continue to Claude Cert Academy