CCDV-F Domain Blueprint
The eight Developer domains, their exam weights, and what each one really tests — at a glance.
Study time should follow the weights. Applications & Integration alone is a third of the exam — build a real app against the API before anything else. Model Selection and Agents & Workflows are the next-biggest technical domains; Claude Code and Eval/Debugging are small but still worth the easy points.
V1
14.7%Agents and Workflows
V2
33.1%Applications and Integration
V3
3.1%Claude Code
V4
2.6%Eval, Testing, and Debugging
V5
16.8%Model Selection and Optimization
V6
11%Prompt and Context Engineering
V7
8.1%Security and Safety
V8
10.6%Tools and MCPs
What each domain tests
| Domain | Wt | The core question it asks |
|---|---|---|
| V1 Agents & Workflows | 14.7% | Workflow, agent, or single call — and can you build it with the SDK? |
| V2 Applications & Integration | 33.1% | Can you correctly implement the API, and ship it like a real engineer? |
| V3 Claude Code | 3.1% | Which config mechanism belongs at which scope? |
| V4 Eval, Testing & Debugging | 2.6% | Which layer actually failed — API, tool, prompt, or model? |
| V5 Model Selection & Optimization | 16.8% | Right model tier, right cost lever, for this exact constraint |
| V6 Prompt & Context Engineering | 11.0% | Structure and budget context — don't just add more of it |
| V7 Security & Safety | 8.1% | Is this enforced in code, or just requested in a prompt? |
| V8 Tools & MCPs | 10.6% | Built-in tool, custom tool, Skill, or MCP server — which fits? |
The meta-question
CCDV-F rewards implementation correctness, not architectural taste. If two options both 'work,' the exam wants the one that matches documented API/SDK behavior exactly — not the more elegant-sounding one.
All CCDV-F references
Independent, unofficial study material from Claude Cert Prep. Not affiliated with Anthropic.