CCDV-F Study Plan: 4-Week Developer Prep Guide

A solid ccdv-f study plan begins with one number: 33.1. That is the percentage of the Claude Certified Developer, Foundations exam (CCDV-F) occupied by Domain 2, Applications and Integration, and it tells you immediately where to spend the majority of your revision time. The exam runs 53 items in 120 minutes, costs $125 USD per attempt, and is scored on a 100-to-1000 scale with 720 as the passing mark. What follows is a structured approach to covering all eight domains efficiently, with study-order rationale grounded in the official domain weights.

What is the CCDV-F exam format, and why does it matter for your study plan?

The CCDV-F is the developer track of Anthropic's Foundations certification programme, delivered online-proctored or at a Pearson VUE test centre. Unlike the Architect track (CCAR-F), the CCDV-F has no scenario bank: items are written directly against the skills in each domain, which means every sitting draws from the same domain coverage. The practical implication is that you cannot rely on spotting scenario patterns across attempts; you need consistent command of each domain's underlying skills.

At 53 items and a 120-minute limit, you have just under two and a half minutes per question on average. The exam uses multiple-choice and multiple-response items; items that require more than one answer state how many to select. The credential is valid for 12 months from the award date, so timing your sitting relative to a project or career milestone is worth considering if renewal cost matters.

What do the CCDV-F domain weights tell you about study priorities?

The eight domains carry precise, published weights per Anthropic's official CCDV-F exam guide (2026-07-08):

Domains 2, 5, and 1 together account for roughly 65% of the exam. A CCDV-F study plan that does not front-load these three domains will struggle to clear 720. Domains 3 and 4 together carry under 6%, but they remain testable: neglecting them entirely risks dropping marks in an area where a few hours of focused study is sufficient.

How should you sequence your CCDV-F study plan week by week?

The structure below assumes four weeks of preparation at eight to ten hours per week. Developers already shipping Claude integrations may compress to three weeks; candidates newer to the API should budget five.

Week 1: Applications and Integration (Domain 2) and Model Selection (Domain 5)

Domain 2 represents a third of the exam. Start here. Cover the Messages API request-response cycle, streaming responses, batch processing constraints, and integration patterns for external systems. Then move to Domain 5: study how to choose between Haiku, Sonnet, and Opus-class models on cost, latency, and capability grounds. Practice constructing scenarios where the correct model choice changes based on the dominant constraint.

Week 2: Agents and Workflows (Domain 1) and Tools and MCPs (Domain 8)

Domain 1 covers agentic loop mechanics and common anti-patterns, hub-and-spoke orchestration, subagent delegation, and error routing between agents. Domain 8 covers MCP integration patterns, tool description design, and the semantics of the isError flag. These two domains share conceptual ground and study well in tandem.

Week 3: Prompt and Context Engineering (Domain 6) and Security and Safety (Domain 7)

Domain 6 tests practical prompt construction: few-shot example design, system prompt structure, and managing context under token constraints. Domain 7 examines prompt injection defences, least-privilege tool scoping, and how to handle ambiguous or sensitive requests. Neither domain is large, but both require scenario-level application rather than definition recall.

Week 4: Claude Code (Domain 3), Evals and Testing (Domain 4), and timed practice

Spend the first part of the week on the two lightest domains, then shift to timed full-length practice exams scored 100 to 1000. Review per-domain percent-correct on each attempt. If Domain 2 or Domain 5 scores are below 70%, return to those materials before your sitting date rather than continuing with new practice runs.

What hands-on projects should you build for CCDV-F preparation?

Exam items test practical judgment. Developers who have built something real in each major domain answer with greater confidence and speed. We suggest three focused builds:

1. A streaming chat application. Call the Messages API with streaming enabled, handle delta events, and render output progressively. This directly covers the integration patterns that Domain 2 tests most frequently.

2. A model-routing layer. Implement a function that accepts a request and routes it to Haiku, Sonnet, or Opus based on token budget, latency target, and task complexity. This operationalises Domain 5 content in a way that multiple-choice revision alone does not.

3. A minimal MCP server. Implement a server with two or three tools, write descriptions that disambiguate routing, and integrate it with a Claude agent. MCP server integration best practices cover the patterns that Domain 8 items probe. Focus on tool description clarity and the distinction between a valid empty result and an access failure, as this distinction appears repeatedly in Domain 8 scenarios.

Each project takes a few hours to reach a functional state. The goal is not production-grade code; it is the decision-making experience the exam surfaces through practical items.

How does CCDV-F test MCP, tool use, and integration patterns?

Domain 8 (Tools and MCPs, 10.6%) appears in integration scenarios that also intersect Domain 2. The exam tests whether you can identify correct tool descriptions, diagnose misrouting caused by ambiguous descriptions, and choose between tool_choice configuration options. The isError flag is a recurring test point: knowing when to set it (an access failure) versus when to return empty content (a valid but empty result) matters for how a downstream agent behaves.

Domain 2 integration items often require you to identify the correct parameter combination for a given requirement. Understanding why the following fields are present, and what changes if any are removed, is the kind of judgment the exam tests:

response = client.messages.create(
    model="claude-sonnet-5",
    max_tokens=1024,
    stream=True,
    messages=[{"role": "user", "content": "Summarise this report."}],
)

Knowing what stop_reason values indicate and how they affect response handling covers ground tested in both Domain 2 and Domain 1.

What does the exam style mean for prompt and context engineering prep?

Domain 6 (Prompt and Context Engineering, 11.0%) rewards developers who understand why a technique works, not just that it exists. Items tend to present a scenario with a failing or underperforming prompt and ask which single change produces the most targeted improvement.

Prompt engineering study should cover three areas: few-shot example construction (when examples help versus when they introduce noise or constrain output format in undesirable ways), system prompt structure (how constraints interact with user instructions), and context management (how to handle conversations approaching the token limit without degrading output quality).

Avoid the pattern of studying definitions in isolation. Work through the Messages API request-response cycle as an end-to-end flow: how the prompt reaches the model, how the model processes it, and how stop conditions affect what is returned. This end-to-end perspective connects Domain 2 mechanics with Domain 6 prompt design.

How much attention should security, evals, and Claude Code receive?

Domain 7 (Security and Safety, 8.1%) carries roughly three times the combined weight of Domains 3 and 4. The primary topics are prompt injection defences, safe tool scoping, and handling requests that touch safety boundaries. The exam rewards proportionate, least-privilege designs rather than blanket restrictions.

Domains 3 (Claude Code, 3.1%) and 4 (Eval, Testing, and Debugging, 2.6%) together represent under 6% of the exam. For Claude Code, focus on the CLAUDE.md configuration hierarchy and hook-based automation. For evals, understand how to design a test that reliably distinguishes a genuine model failure from environmental noise or prompt ambiguity. These areas are testable and focused study is worthwhile; they simply do not justify the same preparation hours as Domains 2 or 5.

A rough time allocation calibrated to the domain weights:

How does AI Skill Certs support a CCDV-F study plan?

AI Skill Certs is an independent adaptive prep platform and is not affiliated with or endorsed by Anthropic. The platform offers adaptive study, Archie Socratic tutoring, and practice exams for CCDV-F. Practice exams mirror the real format: 53 questions scored 100 to 1000 with 720 as the passing bar, with a per-domain percent-correct breakdown mapped to the eight official domains.

The adaptive engine uses Bayesian Knowledge Tracing with a 0.90 mastery threshold, which means the platform continues presenting items in a domain until your modelled proficiency meets that bar. For a CCDV-F study plan, this matters most in Domain 2: a domain that represents a third of the exam benefits from targeted repetition rather than early sign-off.

CCDV-F adaptive study and practice exams are live on the platform today. Note that the CCDV-F concept library at /concepts is not yet available; the /concepts library currently covers the CCAR-F architect track only.