These notes are the reading part of your preparation. Each lesson takes five to seven minutes, starts with a 30-second summary, explains one idea with a diagram and a concrete example, lists the traps exam questions are built around, and ends with two practice questions. Read them in order the first time; come back to single lessons when a practice result shows a weak domain.
How to use the notes
Read one or two lessons a day alongside the four-week study plan. After each lesson, try the two check-yourself questions before opening the answers. When you have read every lesson in a domain, take the diagnostic or a full practice exam and return to the lessons behind any question you missed.
The notes explain the ideas behind the exam's task statements in our own words. They are not a copy of Anthropic's materials: for the authoritative scope, read the official exam guide, and for exact API behaviour, Anthropic's documentation. Each lesson notes when it was last checked.
All lessons
Domain 1: Agentic Architecture & Orchestration (27%)
- The agentic loop: how Claude uses tools6 min · A visual, plain-English explanation of the tool-use loop: stop reasons, tool_result blocks, loop limits and what your code is responsible for.
- Coordinators and subagents, explained6 min · When multi-agent designs pay off, how to brief a subagent, what it should return, and how to handle parallel work and conflicting findings.
- Where should a rule live? Prompts vs code vs permissions5 min · A visual guide to enforcing rules in Claude systems: what belongs in prompts, in handler checks and hooks, in permissions, and with a human.
- Escalation and handoff: when an agent should pass to a human5 min · Escalation triggers, clarifying ambiguous requests, and what a useful handoff packet contains — illustrated for customer support agents.
Domain 2: Tool Design & MCP Integration (18%)
- Designing tools Claude can use well6 min · How to write tool names, descriptions and parameters Claude chooses correctly, keep results small, and return errors an agent can recover from.
- MCP in one picture: servers, primitives and transports5 min · A plain-English, visual explanation of the Model Context Protocol: tools, resources and prompts, stdio versus Streamable HTTP, and securing shared servers.
Domain 3: Claude Code Configuration & Workflows (20%)
- A map of Claude Code configuration7 min · Where every Claude Code setting lives — CLAUDE.md, path rules, settings and permissions, commands, skills, subagents, hooks and MCP — shown as one map.
- Claude Code in CI: headless runs that are safe and parseable5 min · How to run Claude Code in pipelines: print mode, restricting tools, schema-validated output, independent review passes and batch jobs, with a diagram.
Domain 4: Prompt Engineering & Structured Output (20%)
- Prompt structure that produces consistent results6 min · A visual guide to production prompts for Claude: system prompts, explicit criteria, few-shot examples, XML tags, reasoning and measuring changes.
- Structured output and the validation-retry loop6 min · How to get reliable JSON from Claude: structured outputs, strict tools, tool_choice and its limits, honest schemas, truncation checks and capped retries.
Domain 5: Context Management & Reliability (15%)
- Prompt caching and the context window6 min · How prompt caching reuses an identical prefix, why one timestamp breaks it, and how to keep long conversations and large inputs inside the context window.
- Reliability: retries, human review and provenance6 min · Rate limits and backoff, circuit breakers, idempotent batch jobs, routing work to human review, and keeping sources visible in multi-source answers.