Exam Prep: Claude Certified Developer (CCDV-F)

Module 1: Architecting AI-Enabled Applications

• Choosing between workflows, single agent, and agent/subagent designs
• Building agents with the Claude Agent SDK, custom agent loops and harnesses
• Choosing deployment strategies: Self-hosted vs. Anthropic-hosted
• Implement agent design patterns: Tool-use loops, subagents, memory, and context-window management
• Selecting among agentic abstraction frameworks (Strands, LangGraph, PydanticAI)
• Designing applications for Claude Code, Claude Desktop, claude.ai, Claude API, and Claude SDKs

Module 2: Model Selection and Optimisation

• Understanding LLMs, tokens, context windows, long with the impact of non-determinism, and next-token generation
• Selecting between fast mode, extended thinking, and adaptive thinking modes
• Working with effort levels
• Deciding when to use zero-shot, single-shot, and multi-shot prompting
• Picking the right model of Opus, Sonnet, and Haiku
• Exploiting cost management techniques beginning with token usage tracking, cost modeling, and caching

Module 3: Integrating Claude into Applications

• Turning application requirements into AI solutions
• Building with messages, tools, streaming, vision, and thinking
• Configuring Claude system components, including CLAUDE.md files, settings.json, model version pinning, prompt versioning, and plugin dependencies
• Implementing error handling, including effective recovery strategies
• Debugging with trace analysis to identify failure modes and isolate problems
• Integrating REST and Web sockets APIs
• Combing Claude’s Messages API with batch API to reduce costs

Module 4: Prompt and Context Engineering

• Managing the context window
• Preventing context drift and bloat with output pruning and compaction
• Isolating context with subagents and multi-step workflows
• Engineering prompt: Principles and methods
• Improving processing with system prompts. user prompts, and constrained output
• Handling output with established patterns and techniques to produce, validate, and consume output
• Increasing output confidence through structured output patterns, response validation, defensive parsing, and skepticism

Module 5: Accelerating Development with Claude Code

• Exploiting Claude Code Rules, Skills, Commands, Agents, Agent Memory, and session management
• Creating built-in and custom slash commands
• Operating in headless, streaming mode, and auto-mode

Module 6: Extending Applications with Tools and MCP Servers

• Understanding the tradeoffs among built-in tools, custom tools, Skills, and MCP servers
• Best practices for constructing tools and toolsets
• Writing effective tool descriptions
• Picking the right tool usage pattern (agentic harness dispatch, selecting between client-side and server-side tools, choosing approval patterns)
• Configuring for external system interaction
• Best practices in MCP server development and communication when integrating external resources

Module 7: Security and Safety

• Ensuring data privacy and applying security best practices
• Mitigating prompt injection and sanitising prompts
• Defending against jailbreaks
• Ensuring authentication, authorisation, confidentiality, privacy, and integrity.
• Implementing safe and responsible deployment practices with content policy and guardrail layering
• Implementing secure-by-design principles, including using hooks for guardrails and other safety controls
• Managing secrets, credentials, and API keys across development and production environments