CCAR-F
Claude Certified Architect – Foundations: Independent Exam Prep
Independent preparation for the Claude Certified Architect – Foundations (CCAR-F) exam that assumes no AI background. It starts with what Claude is, how programs call it, tools and agents, then covers each exam domain from the basic idea to exam-level judgement: agentic architecture and orchestration, prompt engineering and structured output, tool design and MCP, context management and reliability, and Claude Code configuration and CI/CD, each with worked scenarios and practice questions. Scenario clinics and mock exams are being added. Not affiliated with or endorsed by Anthropic. Anthropic opens exam registration only to people at Claude Partner Network organizations.
DigiTrust course7 modules
Free, access on approval
Modules
M1Foundations from Zero: Claude, the API, Tools and Agents
- ·What Claude is: language models, tokens and the context window25 min
- ·Talking to Claude from code: the Messages API30 min
- ·Giving Claude good instructions: prompting basics20 min
- ·Tools: letting Claude look things up and take actions30 min
- ·From chatbot to agent: the loop25 min
- ·The Claude toolbox: apps, API, Claude Code, Agent SDK and MCP20 min
- ·Ready reckoner10 min
M2The CCAR-F Exam and How to Prepare
- ·How the CCAR-F exam works and who can take it15 min
- ·Reading scenario questions and spotting distractors20 min
- ·Your study plan and lab setup15 min
- ·Ready reckoner10 min
M3Agentic Architecture & Orchestration27%
- ·Choosing the right pattern: from single calls to agents30 min
- ·Putting brakes on the agent loop30 min
- ·A lead agent and its helpers: coordinators and subagents30 min
- ·Splitting a big task well: task decomposition20 min
- ·Knowing when to hand over to a human: escalation20 min
- ·Keeping the agent's power small: narrow write surfaces20 min
- ·Worked scenario: a customer support resolution agent25 min
- ·Ready reckoner10 min
M4Prompt Engineering & Structured Output20%
- ·System prompts for production30 min
- ·Examples and reasoning: few-shot and step-by-step thinking30 min
- ·Structured output: getting data your code can trust35 min
- ·Validate, retry, then hand over: making output reliable25 min
- ·Worked scenario: structured data extraction at scale30 min
- ·Ready reckoner10 min
M5Tool Design & MCP Integration18%
- ·Designing tools Claude uses well30 min
- ·Tool errors, permissions and composition30 min
- ·How MCP works: hosts, clients, servers and transports30 min
- ·Building and securing an MCP server30 min
- ·Worked scenario: developer productivity with Claude and MCP25 min
- ·Ready reckoner10 min
M6Context Management & Reliability15%
- ·Managing the context window in long tasks30 min
- ·Prompt caching: paying less for what doesn't change25 min
- ·Rate limits, retries and timeouts25 min
- ·Circuit breakers and graceful degradation20 min
- ·Provenance: answers you can trace and trust25 min
- ·Worked scenario: making a multi-agent research system reliable25 min
- ·Ready reckoner10 min
M7Claude Code Configuration & Workflows20%
- ·Claude Code and CLAUDE.md: giving the agent standing instructions30 min
- ·Settings, permissions and plan mode30 min
- ·Hooks: rules that run every time25 min
- ·Custom commands, skills and subagents in Claude Code25 min
- ·Claude Code without a person watching: headless runs and CI/CD30 min
- ·Worked scenarios: code generation with Claude Code and Claude Code in CI/CD30 min
- ·Ready reckoner10 min