Resources
The official material behind every lesson here. If a lesson and a document disagree, the document wins; tell us and we fix the lesson.
Official courses
Anthropic Academy courses are the closest thing to the exam's source material. All are self-paced.
- Anthropic AcademyCatalogue of every course, plus certification information and booking.
- Claude 101Foundations for everyday use: prompting, Projects, Artifacts. Core for CCAO-F.
- Building with the Claude APIMessages API, tool use, streaming, batches. Core for CCDV-F and CCAR-F.
- Introduction to Model Context ProtocolHosts, clients, servers, transports. Tested on every technical exam.
- Claude Code in ActionSettings, hooks, permissions, subagents, CI. Twenty percent of CCAR-F.
Documentation
Read the pages the lessons link to. Exam questions are written from these.
- Claude Developer Platform docsAPI reference, guides, model overview, pricing, rate limits.
- Prompt engineering guideAnthropic's own prompting techniques, in the order they recommend trying them.
- Prompt cachingBreakpoints, minimum lengths, TTL, what invalidates a cache.
- Tool useSchemas, the tool_use and tool_result cycle, parallel calls, tool_choice.
- Claude Code docsCLAUDE.md, settings hierarchy, hooks, subagents, skills, MCP, headless mode.
- Claude Agent SDKBuilding production agents on the same harness Claude Code uses.
- Model Context Protocol specificationThe spec, SDKs and reference servers.
- Usage policyWhat may and may not be built. Governance questions draw on this.
Engineering writing
Longer reads that explain the reasoning behind the patterns the exams test.
- Building effective agentsWorkflows versus agents, and the five composable patterns. The single most-cited source for architecture questions.
- Building agents with the Claude Agent SDKThe agent loop, context management, tooling and verification.
- Effective context engineering for AI agentsWhy context is finite and how to spend it.
- Writing tools for agentsTool design guidance: naming, descriptions, consolidation, error messages.
- Anthropic engineering blogEverything else, including multi-agent research systems and Claude Code best practices.
Code
Working examples to run and change.
- Anthropic CookbookNotebooks for tool use, RAG, caching, vision, agents, evals.
- Claude Agent SDK, PythonReference implementation and examples.
- Claude Agent SDK, TypeScriptSame, for Node.
- MCP reference serversFilesystem, git, fetch and other servers to read for structure.
- Claude Code repositoryIssues, changelog and examples of hooks and settings.