Claude Certified Developer-Foundations (CCDV-F) - 6 Exams

This practice test bundle contains multiple full-length practice exams for the Anthropic Claude Certified Developer — Foundations (CCDV-F) exam. Each exam is built from a pool of original questions, weighted by domain to match the official exam blueprint.

D1: Agents and Workflows (14% of the exam), covering workflow vs. agent: decision criteria and tradeoffs, manager/supervisor hierarchies and the role of subagents, claude agent sdk: query(), the agent loop, and input modes, custom agent loops and harnesses, managed agent deployment: self-hosted vs. anthropic-hosted, hooks for deterministic actions, agent design patterns: tool-use loops, subagents, memory, context-window management, agentic frameworks: langgraph vs. strands vs. pydanticai (identities and tradeoffs). You will understand how each of these areas is tested on the exam and how they connect to real-world practice.

D2: Applications and Integration (33% of the exam), covering deriving functional requirements from business needs, infrastructure requirements and solution-architecture fit, systems life cycle: develop, implement, operate, maintain, messages api structure and data-access patterns, tools, streaming, and vision, extended/adaptive thinking and prompt caching, message batches api: realtime vs. batch tradeoffs, invoking claude via third-party vendors (bedrock, vertex), rest apis and json, asynchronous programming patterns, version control and sdlc integration, code review and refactoring, instruction interpretation across interfaces (claude code / desktop / claudeai / api / sdks), content boundaries and schema design, session hygiene, plugin management, claudemd files and hierarchy, settingsjson configuration, model version pinning and prompt versioning, plugin dependencies. You will understand how each of these areas is tested on the exam and how they connect to real-world practice.

D3: Claude Code (3% of the exam), covering core components: rules, skills, commands, agents, agent memory, session management and slash commands (built-in + custom), headless / streaming / auto modes, claudemd hierarchy, repo init, settingsjson. You will understand how each of these areas is tested on the exam and how they connect to real-world practice.

D4: Eval, Testing, and Debugging (2% of the exam), covering error type identification and recovery strategy selection, trace analysis and problem-origin isolation (integration layer vs. model output). You will understand how each of these areas is tested on the exam and how they connect to real-world practice.

D5: Model Selection and Optimization (16% of the exam), covering llm basics: tokens, context windows, sampling, non-determinism, next-token generation, model options: fast mode, extended/adaptive thinking, effort levels, fundamental prompting: zero-, single-, multi-shot, sdks that wrap rest apis, websockets and connection fundamentals, opus vs. sonnet vs. haiku: use cases and quality/latency/cost tradeoffs, breaking behavior changes across model releases, token usage tracking and cost modeling, caching for cost: prompt caching and cache check-pointing. You will understand how each of these areas is tested on the exam and how they connect to real-world practice.

D6: Prompt and Context Engineering (11% of the exam), covering context window management and drift/bloat prevention (tool-output pruning, compaction), context isolation via subagents and multi-step workflows, instruction clarity, system vs. user placement, output constraints, few-shot examples and iterative refinement, input sanitization, structured output patterns and response validation, defensive parsing and skepticism toward confident output. You will understand how each of these areas is tested on the exam and how they connect to real-world practice.

D7: Security and Safety (8% of the exam), covering prompt injection and jailbreak defense, untrusted input, data-leakage prevention, and pii handling, content policy and guardrail layering, secure-by-design: privacy, iam, least privilege, hooks as guardrails: preventing destructive actions, managing secrets, credentials, and api keys (dev + prod), identity validation, access approval, and authorized-access monitoring. You will understand how each of these areas is tested on the exam and how they connect to real-world practice.

D8: Tools and MCPs (10% of the exam), covering tool use and function calling: descriptions and schemas, tool error handling and usage patterns (agentic dispatch, client vs. server, approval), tool set construction best practices, mcp server authoring: resources, tools, prompts, mcp communication patterns: stdio, sockets, client vs. server, choosing among built-in tools, custom tools, skills, and mcps (tradeoffs). You will understand how each of these areas is tested on the exam and how they connect to real-world practice.

Exams are arranged in graduated difficulty, from warm-up through exam simulation, so you can build confidence before testing under real conditions. Every question includes a detailed explanation covering why the correct answer is right and why each distractor is wrong.

Major topics covered: workflow vs. agent: decision criteria and tradeoffs, manager/supervisor hierarchies and the role of subagents, claude agent sdk: query(), the agent loop, and input modes, custom agent loops and harnesses, managed agent deployment: self-hosted vs. anthropic-hosted, hooks for deterministic actions, agent design patterns: tool-use loops, subagents, memory, context-window management, agentic frameworks: langgraph vs. strands vs. pydanticai (identities and tradeoffs), deriving functional requirements from business needs, infrastructure requirements and solution-architecture fit, systems life cycle: develop, implement, operate, maintain, messages api structure and data-access patterns, tools, streaming, and vision, extended/adaptive thinking and prompt caching, message batches api: realtime vs. batch tradeoffs, invoking claude via third-party vendors (bedrock, vertex), rest apis and json, asynchronous programming patterns, version control and sdlc integration, code review and refactoring, instruction interpretation across interfaces (claude code / desktop / claudeai / api / sdks), content boundaries and schema design, session hygiene, plugin management, claudemd files and hierarchy, settingsjson configuration, model version pinning and prompt versioning, plugin dependencies, core components: rules, skills, commands, agents, agent memory, session management and slash commands (built-in + custom), CCDV-F exam prep 2026.