CCDV-F mock exams
Several independent full-length attempts. Every question is a judgment call — plausible, equal-length options where you pick the best decision, not the only correct one. Take them fresh, review every explanation, and use the per-domain scorecard to find what to study next.
Mock 1 · Comprehensive
28 questions · all 8 domains
Weight-proportional coverage across all eight domains — Claude API mechanics, model selection and cost, agents vs. workflows, prompt/context engineering, tools and MCP (including the July 2026 spec revision), security, Claude Code, and debugging.
Mock 2 · Comprehensive
22 questions · all 8 domains
A second independent full attempt with fresh scenarios across all eight domains — truncation and refusal handling, spend-cap errors, tool-choice forcing, direct prompt injection, and diagnostic use of stop_reason.
Mock 3 · Comprehensive
22 questions · all 8 domains
A third independent full attempt — Batches API ceilings, model-tier selection under a capability requirement, MCP primitive selection, headless CI usage, least-privilege tool scoping, and regression-testing discipline for prompt changes.
Mock 4 · Level 3 Advanced
53 questions · all 8 domains
The hardest attempt: high-criticality production scenarios (duplicate charges, cross-tenant data bleed, prompt injection, runaway agent loops) built around root-cause and layered reasoning rather than recall — audited for zero answer-position, length, or keyword bias.