The Claude Certified Developer (Foundations) exam, code CCDV-F, is the hands-on builder credential in Anthropic's certification programme. It tests whether you can build, integrate, and ship production-grade applications, agents, and workflows on the Claude platform at a foundational level. Where the Associate exam is about using Claude and the Architect exam is about designing systems, Developer is about writing the code that turns Claude's capabilities into shipped software.
The full prep track for this exam is live here: 25 lessons, one per weighted sub-skill in the exam guide, a free diagnostic, and six full-length mock exams. Below is the published blueprint. Read it closely. It's the most granular of the four exams, and the weightings are unusually skewed.
Who it's for
The target candidate is a technical builder: AI and ML engineers, technical leads, and senior software engineers. It is explicitly not for non-technical or casual users, nor for anyone whose experience with Claude is limited to writing prompts.
Recommended background is concrete: one to five years in software engineering, at least six months hands-on with Claude or comparable LLM systems, proficiency in Python and/or TypeScript, fluency with REST APIs and command-line tools, and a working understanding of LLM fundamentals, agents, context management, and MCP. As with every exam in the programme, that experience is recommended, not required, and the credential is awarded on exam performance alone.
The eight domains, and a very uneven blueprint
The Developer exam uses a two-level blueprint: eight domains broken into 25 individually weighted sub-skills. What stands out is how much weight sits in one place.
| Domain | Weight | Focus |
|---|---|---|
| Applications and Integration | 33.1% | API mechanics, application design, software-engineering foundations, config management |
| Model Selection and Optimisation | 16.8% | LLM and technical fundamentals, model trade-offs, cost and token management |
| Agents and Workflows | 14.7% | Agent architecture, construction with Claude, patterns and frameworks |
| Prompt and Context Engineering | 11.0% | Context engineering, prompt engineering, output handling |
| Tools and MCPs | 10.6% | Tool implementation, MCP server development, agentic customisation |
| Security and Safety | 8.1% | Application security, guardrails, hooks, secrets and key management |
| Claude Code | 3.1% | Operating Claude Code |
| Eval, Testing, and Debugging | 2.6% | Debugging and error handling |
Applications and Integration is a third of the entire exam on its own. Its sub-skills, Claude Application Design at 8.6%, Software Engineering Foundations at 7.4%, and Claude API Mechanics at 6.8%, are three of the heaviest single items on the whole blueprint. The message is unambiguous: this exam rewards people who can actually stand up a well-engineered application against the Claude API, not people who can only describe one.
At the other end, Claude Code and debugging are light, at 3.1% and 2.6%. They're on the exam, but they're not where it's decided. If you're budgeting study time, the ratio between the top and bottom of that table should shape it directly.
Where it sits among the four certifications
Developer is the builder credential in a programme of four Claude certifications:
- Associate uses Claude as a productivity tool, no code required.
- Developer builds and ships Claude-powered applications through the API, Agent SDK, Claude Code, and MCP.
- Architect designs those systems end to end and owns the architectural trade-offs.
At 53 items it has the fewest questions of the four, and like the rest it runs 120 minutes, needs a scaled 720 out of 1,000 to pass, and stays valid for 12 months. There's no prerequisite; if you build with Claude for a living, you can sit it directly.
How to use the prep track
Take the free diagnostic first to find your weak domains, then work through the curriculum in blueprint order. Budget your time by weight. Applications and Integration gets six lessons and a third of the marks, while Claude Code and debugging get one lesson each. Every lesson is free. When you're ready, sit a mock exam: 53 questions, 120 minutes, and the same 720 pass mark as the real thing.
This is an independent community resource, not affiliated with or endorsed by Anthropic. Registration for the official exam runs through Anthropic's Partner Academy, delivered by Pearson VUE.