The Claude Certified Developer – Foundations (CCDV-F) exam.

What is CCDV-F?

CCDV-F (Claude Certified Developer – Foundations) is Anthropic's certification for developers who build applications on Claude. The exam is 53 questions in 120 minutes, proctored and closed-book, delivered by Pearson VUE, with a passing score of 720 on a 100–1,000 scale and a $125 fee per attempt.

CCDV-F at a glance

What the exam covers

Eight domains, per exam guide v1.0 (July 2026).

Try three CCDV-F practice questions

One from each of the heaviest domains, with the answer and the reasoning. Every question in our bank is original, written from the published blueprint, never taken from the exam.

More than 250 people have started the free CCDV-F practice exam since September, and more than 180 finished a full timed mock. More than 220 engineers have joined the YAID Network. Counts as of October 15, 2026. Keep going with all 60 free CCDV-F practice questions or the timed CCDV-F mock.

Who it’s for

The exam guide describes the target candidate precisely: an engineer with one to five years of experience, at least six months hands-on with Claude or comparable LLM systems, proficient in Python or TypeScript, and fluent with REST APIs and CLI tools. A third of the exam is applications and integration, so it rewards people who have shipped.

How CCDV-F compares

Against the rest of the program: CCAR-F tests system design where CCDV-F tests the code itself, CCAO-F is the non-developer track, and CCAR-P extends the architect side. Developers who spend their days shipping Claude features fit here; a third of the marks are applications and integration, which no other exam in the program weights so heavily.

How to register

Registration runs through the Anthropic Partner Academy, and access is open to organizations in the Claude Partner Network. There is no individual sign-up. If your employer is not a partner, the practical route is joining a partner firm: YAID members register under our partner standing, prepare with the official Partner Academy materials, and sit the exam through Pearson VUE. The credential is issued to you personally and stays with you. The full path is in the Claude certification guide.

Preparing for CCDV-F

1. Applications and integration is a third of the exam. If your Claude experience is mostly prompting, build and deploy something real against the API before you sit it.

2. Claude Code and eval tooling together are under 6% of the marks. Do not over-prepare the small domains at the expense of integration and model selection.

3. The guide's experience profile is honest: six months of hands-on Claude work is the real prerequisite, even though no formal one is enforced.

Test yourself with the free timed CCDV-F practice exam or the CCDV-F practice questions by domain, both written from the blueprint and weighted the way the exam is. For a week-by-week schedule, read a four-week CCDV-F plan built on the blueprint. To revise in short bursts, use the CCDV-F flashcards.

What each domain covers

What to know before you sit, organized by the published blueprint. Nothing here describes an exam question.

1. Applications and integration

   • The Messages API: request shape, message roles, system prompt placement, max_tokens, and what each stop_reason means (end_turn, max_tokens, tool_use, stop_sequence).
   • Streaming responses and how to handle partial content and tool-use blocks as they arrive.
   • Multimodal inputs: images and PDFs as content blocks, and the Files API for reusing uploads.
   • Structured output: JSON via tool definitions or structured outputs, and validating what comes back.
   • The Batch API for high-volume, non-urgent work at lower cost.
   • Errors, rate limits, and retries: status codes, exponential backoff, and idempotent design for anything that writes.
   • Prompt caching mechanics: cache breakpoints, minimum lengths, and ordering stable content first.

2. Model selection and optimization

   • The model family trade-off: the most capable model for hard reasoning, the mid-tier for most production work, the smallest for high-volume classification and routing.
   • Cost is tokens in plus tokens out; latency is dominated by output tokens. Optimize both by trimming prompts, capping max_tokens, and caching.
   • Extended thinking budgets: when deeper reasoning pays for itself and when it is wasted.
   • Temperature and sampling settings, and why deterministic tasks want low temperature.
   • Routing patterns: a small model classifies, a larger model handles the hard cases.

3. Agents and workflows

   • Implementing the agent loop in code: send messages, detect tool_use blocks, execute, append tool_result blocks, repeat until end_turn.
   • Workflow patterns from Anthropic's guidance (chaining, routing, parallelization, orchestrator-workers, evaluator-optimizer) and how to pick one.
   • The Claude Agent SDK: what it handles for you (the loop, tools, permissions, hooks, subagents) versus a hand-rolled loop.
   • Stop conditions, max turns, and human approval steps.

4. Prompt and context engineering

   • System prompt versus user turn, examples, XML structure, prefilling, and long-context ordering.
   • Managing conversation history: truncation, summarization, and what to keep verbatim.

5. Tools and MCPs

   • Tool definitions with JSON schema, descriptions written for the model, tool_choice, parallel tool calls, and error results.
   • Connecting to MCP servers from an application and the difference between tools, resources, and prompts.

6. Security and safety

   • Prompt injection through tool results and documents: treat retrieved content as data, and keep instructions in the system prompt.
   • Least privilege for tools: read-only by default, confirmation for writes, no secrets in prompts.
   • Handling personal data, logging, and retention; Anthropic's usage policies as constraints on what you build.
   • Output validation before anything is executed, sent, or stored.

7. Claude Code

   • What Claude Code is, where it runs, CLAUDE.md, and the basics of hooks, MCP configuration, and headless use. A small domain, so keep the prep proportionate.

8. Eval, testing, and debugging

   • Golden test sets, LLM-as-judge grading, regression tests on prompt changes, and logging enough to reproduce a bad response. This is the smallest domain on the exam.

Common mistakes

1. Preparing as if this were the architect exam, when a third of the marks are integration code and Claude Code plus evals together are under 6%.

2. Not knowing the stop_reason values and what to do with each one.

3. Skipping the Batch API, Files API, and prompt caching because they were not needed in your own project.

4. Sitting the exam with mostly prompting experience. The guide asks for six months of hands-on API work and the question mix is built around it.

A study path

1. 01

   Week 1: read the exam guide and the Messages API reference. Build a script that streams a response, calls a tool, and handles every stop_reason.

2. 02

   Week 2: add prompt caching, a batch job, and a PDF input to that script. Measure cost before and after caching.

3. 03

   Week 3: build a small agent loop by hand, then rebuild it with the Agent SDK. Add an eval with a golden set.

4. 04

   Week 4: Partner Academy prep course and a practice exam; closed-book review of the API reference.

Official reading

• Claude Developer Platform docs
• Building effective agents
• Writing tools for agents
• Effective context engineering for AI agents
• Model Context Protocol
• Claude Code docs

The other Claude certifications

Not sure the credential is worth the prep time? The honest case for and against Claude certification.