Claude Certified Developer Exam Guide

About the CCDV-F Certification

The Claude Certified Developer, Foundations certification validates that you can build, integrate and ship production-grade applications, agents and workflows on Anthropic's Claude platform at a foundational level (Section 1 of the exam guide). It covers API integration, agent and tool construction, prompt and context engineering, evaluation, security and model selection.

Section 3 describes the intended audience as AI and machine learning engineers, technical leads and senior software engineers who build, integrate and ship production AI solutions. It is not intended for non-technical users, people without hands-on software development experience, or roles limited to prompt writing.

Choosing between this exam and the Architect exam? See how the CCDV-F compares with the CCAR-F.

CCDV-F Exam Format

All figures are from Section 5 of the exam guide. 53 items in 120 minutes works out to about 2 minutes 15 seconds per item.

Domain Breakdown

The exam covers eight domains and 25 weighted skills (Section 6). The weights are the approximate share of scored items drawn from each domain. Applications and Integration alone is about a third of the exam.

Domain 1: Agents and Workflows (14.7%)

Skills: Agent Architecture (4.5%), Agent Construction with Claude (5.3%), Agent Patterns and Frameworks (4.9%). Covers when to use a workflow versus an agent, manager and supervisor hierarchies, subagents, the Claude Agent SDK, custom agent loops and harnesses, self-hosted versus Anthropic-hosted deployment, hooks for deterministic actions, and agentic frameworks such as Strands, LangGraph and PydanticAI.

Domain 2: Applications and Integration (33.1%)

The largest domain, with six skills: Claude Application Design (8.6%), Software Engineering Foundations (7.4%), Claude API Mechanics (6.8%), Configuration Management (4.1%), Understanding Requirements (3.4%) and Systems Life Cycle (2.8%). Covers messages, tools, streaming, vision, thinking, caching and batch use in the Claude API; how Claude interprets instructions across Claude Code, Desktop, claude.ai, the API and SDKs; schema design and session hygiene; CLAUDE.md, settings.json, model version pinning and prompt versioning; and core engineering practice such as REST APIs, JSON, asynchronous programming, version control and code review.

Domain 3: Claude Code (3.1%)

Skill: Claude Code Operation (3.1%). Covers Rules, Skills, Commands, Agents and Agent Memory; session management, built-in and custom slash commands, headless mode, streaming mode and auto-mode; the CLAUDE.md hierarchy, repository initialization and settings.json.

Domain 4: Eval, Testing, and Debugging (2.6%)

Skill: Debugging and Error Handling (2.6%). Covers identifying error types, choosing a recovery strategy, reading traces to find failure modes, and isolating whether a problem comes from the integration layer or the model output.

Domain 5: Model Selection and Optimization (16.8%)

Skills: Technical Fundamentals (6.1%), LLM Fundamentals (5.2%), Cost and Token Management (2.8%), Model Selection and Tradeoffs (2.7%). Covers tokens, context windows, sampling and non-determinism; fast mode, extended thinking, adaptive thinking and effort levels; zero-shot to multi-shot prompting; Opus, Sonnet and Haiku use cases and quality, latency and cost tradeoffs; token tracking, cost modeling and prompt caching; and SDKs that wrap REST APIs and websockets.

Domain 6: Prompt and Context Engineering (11.0%)

Skills: Prompt Engineering (4.6%), Context Engineering (3.8%), Output Handling (2.6%). Covers instruction clarity, few-shot examples, system versus user placement, output constraints and input sanitization; context window management, preventing context drift and bloat through tool output pruning and compaction, and context isolation through subagents; and structured output, response validation and defensive parsing.

Domain 7: Security and Safety (8.1%)

Skills: AI Application Security (3.2%), Guardrails and Safe Deployment (2.3%), Identity, Secrets, and Key Management (1.6%), Claude Hooks (1.0%). Covers prompt injection and jailbreak defense, untrusted input handling, data leakage and PII; guardrail layering, least privilege and secure-by-design principles; managing secrets and API keys across environments; and hooks that prevent destructive actions.

Domain 8: Tools and MCPs (10.6%)

Skills: Tool Implementation (4.4%), Agentic Customization (4.1%), MCP Server Development (2.1%). Covers function calling, tool description writing, tool error handling, client-side versus server-side tools and approval patterns; MCP server authoring and deployment, MCP resources, tools and prompts, and stdio and socket communication; and choosing between built-in tools, custom tools, Skills and MCP servers for a given use case.

Multiple-Response Items

Section 5 of the exam guide says the exam mixes multiple-choice and multiple-response items, and that each item states how many responses to select. The guide does not say how many of the 53 items are multiple-response, so prepare for both formats.

What this means in practice:

• Judge each option on its own. With two answers to pick, there is no single best option to stop at. Read every option as true or false for the situation.
• Select exactly the number stated. The item tells you how many responses it wants.
• Don't count on partial credit. The guide does not describe partial credit, so treat each item as right or wrong.

Our CCDV-F bank includes 44 multiple-response items, and each one states how many answers to select, so you practice the format before exam day.

Study Strategy

Section 7 of the exam guide recommends combining hands-on experience with study. An effective approach:

• Self-assess against the blueprint. Mark each of the 25 skills in Section 6 as strong, partial or new, and start with the new ones in the heaviest domains.
• Build one real application. Section 7 recommends building and operating at least one Claude application that uses the API, integrates one or more tools, and includes basic prompt and context engineering, security and evaluation practices.
• Practice in short sessions. Use Quick Sprint to find weak domains. Without an account it draws from a 28-question free sample; enrolled accounts draw from all 400.
• Simulate real conditions. Take the full 53-question, 120-minute mock at least once before your exam date.

For a week-by-week version, see our domain-by-domain CCDV-F study plan.

Try a Sample CCDV-F Question

Here is a question from our free sample, from the largest domain, Applications and Integration (33.1%):

Sample Question: Applications and Integration

A triage agent calls a route_ticket tool whose priority field is a free-text string. The application switches on that value and keeps falling into its default branch, because Claude sends variants such as High, high priority and urgent. The developer wants values the application can rely on. Which schema change fits best?

• A. Declare priority as an enum of the allowed values in input_schema, so the schema states what the application accepts.
• B. List the allowed values in the tool's description and keep the field a free string, since the description constrains the input.
• C. Change the field to an integer priority code that the application maps to labels, since a number cannot be misspelled.
• D. Add a system prompt instruction requiring the value in uppercase, so that the application needs a single string comparison.

Correct answer: A. The define tools page's example tool expects an input object with a required location string and an optional unit string that must be either celsius or fahrenheit, which is the schema itself stating the allowed values, and the prompting best practices page says that for classification tasks you should use either tools with an enum field containing your valid labels or structured outputs. Why a description alone is wrong: the define tools page says the description explains what the tool does, when to use it and how it behaves, while the input_schema is what defines the expected parameters, so a free string stays free. Why an integer code is wrong: an unconstrained integer is as open as an unconstrained string, and the page's mechanism for a fixed set of values is the enum. Why an uppercase instruction is wrong: it asks the model to follow a convention instead of letting the schema define the accepted values, so the application still cannot rely on what arrives.

That's one of 400 questions across all eight domains in our CCDV-F practice questions. Every one explains the correct answer and why each other option is wrong.

Study Timeline

How long you need depends on how much of the recommended background you already have. Section 4 recommends one to five years of software engineering experience, at least six months of hands-on work with Claude or a comparable LLM system, proficiency in Python or TypeScript, and fluency with REST APIs and command line tools.

If you build with Claude every day, a focused week or two of practice, followed by a timed mock, may be enough. If you are newer to it, give yourself about four weeks and cover the domains in weight order. Our four-week CCDV-F study plan maps each week to the skills that carry the most weight.

Common Mistakes to Avoid

• Skipping Claude Code because its domain is only 3.1%. CLAUDE.md and settings.json also sit under Configuration Management in the largest domain.
• Ignoring the general engineering skills. Software Engineering Foundations, Technical Fundamentals, Understanding Requirements and Systems Life Cycle together carry 19.7% of the exam.
• Answering multiple-response items like single-choice ones. Select exactly the number of responses the item asks for.
• Treating the guide's sample questions as the exam. Section 8 says they are illustrative and not drawn from the live item bank.
• Never practicing against the clock. 53 items in 120 minutes leaves about 2 minutes 15 seconds each.

Test Your Readiness

Start with the free 10-question CCDV-F Quick Sprint, no account needed. When you are closer to exam day, take the full 53-question mock in 120 minutes, drawn at the guide's domain weights.

How to Register for the CCDV-F Exam

Frequently Asked Questions

How hard is the CCDV-F exam?

Anthropic does not publish pass rates. The exam is written for a candidate with one to five years of software engineering experience and at least six months of hands-on work with Claude or a comparable system (Section 4), and it spans 25 skills across eight domains, about a fifth of them general software engineering.

How much does the CCDV-F exam cost?

$125 USD for the CCDV-F exam, according to Section 5 of the CCDV-F exam guide. The price at checkout reflects any discount for your company's partner tier, and the fee applies to each attempt.

What score do I need to pass the CCDV-F exam?

A scaled score of 720 on a scale of 100 to 1,000. The exam is criterion-referenced, so you are measured against a fixed standard rather than other candidates, and the percent-correct figures by domain on your score report do not decide the result (Section 9).

How long is the CCDV-F exam?

53 items in 120 minutes, about 2 minutes 15 seconds per item. Items are multiple-choice or multiple-response, and each states how many responses to select (Section 5).

How long is the CCDV-F certification valid?

12 months from the date it is awarded. On-time renewal is a free, non-proctored assessment on the Anthropic Partner Academy; if the credential lapses, you retake the full exam at the full fee (Section 14).

Who should take the CCDV-F exam?

Developers who build, integrate and ship Claude-powered applications: AI and machine learning engineers, technical leads and senior software engineers (Section 3). It is not intended for non-technical users, people without hands-on development experience, or roles limited to prompt writing.

Can I retake the CCDV-F exam if I fail?

Yes. The waiting period is 14 days after the first failed attempt, 30 days after the second and 90 days after the third, with up to four attempts in a rolling 12 months and the fee due each time. The limits apply per exam (Section 11). See how retakes and waiting periods work.