Claude Certified Developer – Foundations (exam code CCDV-F) is 53 multiple-choice and multiple-response questions in 120 minutes, across eight domains. The biggest by far is Applications and Integration at 33.1%. It costs $125 before partner discounts, passes at 720 out of 1,000, and you need a Claude Partner Network company email to book it.
Everything in this guide comes from Anthropic's official exam guide (version 1.0, July 2026) and its Partner Academy FAQ, unless I say otherwise. The weightings below are Anthropic's, not estimates.
Exam update on 7 December 2026. Pearson VUE says the Developer – Foundations exam will include updated content from 7 December 2026, and the domain weights below may change. Anthropic has not yet published what changes. This guide reflects the current exam guide (v1.0, July 2026) and will be updated when the new guide is out.
Who is the Claude Certified Developer – Foundations exam for?
It is for engineers who build, integrate and ship Claude applications, agents and workflows in production. Anthropic names AI and ML engineers, technical leads and senior software engineers as the core audience.
The recommended experience, from the exam guide:
- One to five years of software engineering
- At least six months hands-on with Claude or a comparable LLM-based system
- Python and/or TypeScript
- Comfortable with REST APIs and CLI tools
- A working understanding of LLM fundamentals, agents, context management and MCP
That experience is recommended, not required. There are no mandatory prerequisites or courses.
Anthropic is equally clear about who it is not for: non-technical or casual Claude users, and roles limited to prompt writing without broader application development work. If that is you, the Associate is the closer fit. I compare all four in Claude Certified Architect vs Developer vs Associate.
See the study system in action
How StudyTech AI works
Watch how your answers become a focused daily plan, targeted practice and flashcards. See how StudyTech AI tracks your progress before you decide when to book.
Ready to explore? Visit StudyTech AI to see supported certifications and choose your next step.
Explore StudyTech AIExam format, cost and passing score
53 questions, 120 minutes, $125, a scaled pass mark of 720, proctored by Pearson VUE, valid for 12 months.
| Detail | Claude Certified Developer – Foundations |
|---|---|
| Exam code | CCDV-F |
| Questions | 53 |
| Question types | Multiple choice and multiple response; each question states how many answers to select |
| Time | 120 minutes (about 135 minutes total seat time) |
| Price | $125 before partner-tier discounts |
| Partner discount | 50% off for Select, Preferred and Global Premier partners, applied at checkout. Global Premier partners get 100% off through 31 December 2026 |
| Passing score | 720 on a scaled range of 100–1,000 |
| Delivery | Proctored by Pearson VUE, online (OnVUE) or at a test centre |
| ID | Valid, unexpired, government-issued photo ID that matches your registration exactly |
| Result | Pass or fail with a scaled score, plus percent correct by domain |
| Badge | Issued through Credly |
| Validity | 12 months from the date awarded |
| Eligibility | People at Claude Partner Network organisations, with a company-domain email |
| Prerequisites | None |
Three details worth knowing before you book:
- The score report shows your percent correct per domain. If you fail, that tells you exactly where to spend the next study block.
- You can cancel or reschedule up to 24 hours before. Inside 24 hours, or a no-show, and you lose the fee.
- Retakes are strict. Up to four attempts per rolling 12 months, with waits of 14, then 30, then 90 days, and every attempt costs the full fee. The limit is per exam, so failing Developer does not block you from booking a different Claude exam.
The full price and policy breakdown for all four Claude exams is in the Claude certification cost and format guide.
The eight exam domains and their weightings
Applications and Integration is a third of the exam; the three domains after it are Model Selection, Agents and Workflows, and Prompt and Context Engineering.
These are Anthropic's official weightings from the exam guide. They show the approximate share of scored questions drawn from each domain.
| # | Domain | Weight | Rough questions (of 53) |
|---|---|---|---|
| 1 | Agents and Workflows | 14.7% | ~8 |
| 2 | Applications and Integration | 33.1% | ~18 |
| 3 | Claude Code | 3.1% | ~2 |
| 4 | Eval, Testing, and Debugging | 2.6% | ~1 |
| 5 | Model Selection and Optimization | 16.8% | ~9 |
| 6 | Prompt and Context Engineering | 11.0% | ~6 |
| 7 | Security and Safety | 8.1% | ~4 |
| 8 | Tools and MCPs | 10.6% | ~6 |
The "rough questions" column is my own arithmetic (weight × 53), not an Anthropic figure. Anthropic does not say how many of the 53 are scored, so treat it as a guide to effort, not a count.
The surprise in that table: Claude Code is only 3.1%, and evals and debugging only 2.6%. Despite the name, this is an application developer exam first. The API, software engineering fundamentals, application design and configuration carry far more weight than any one tool.
What each domain covers
Each domain is a set of weighted skills; the summaries below follow Anthropic's own skill list, in plain English.
Domain 1: Agents and Workflows (14.7%)
- Agent architecture (4.5%): when to use a fixed workflow versus an agent, manager and supervisor hierarchies, and how subagents improve task execution.
- Agent construction with Claude (5.3%): the Claude Agent SDK, custom agent loops and harnesses, self-hosted versus Anthropic-hosted agents, and hooks for deterministic actions.
- Agent patterns and frameworks (4.9%): tool-use loops, subagents, memory, context-window management, and frameworks such as Strands, LangGraph and PydanticAI.
Domain 2: Applications and Integration (33.1%)
The heavyweight. Six skills:
- Claude API mechanics (6.8%): messages, tools, streaming, vision, thinking, caching, calling Claude through third-party vendors, the batch API, and when to choose realtime versus batch.
- Software engineering foundations (7.4%): REST, JSON, async programming, version control, SDLC integration, code review and refactoring.
- Claude application design (8.6%): how Claude interprets instructions across Claude Code, Desktop, claude.ai, the API and SDKs; content boundaries, schema design, session hygiene and plugin management.
- Configuration management (4.1%): CLAUDE.md files, settings.json, model version pinning, prompt versioning and plugin dependencies.
- Understanding requirements (3.4%) and systems life cycle (2.8%): turning business requirements into functional and infrastructure requirements, and how systems are built, run and maintained.
Domain 3: Claude Code (3.1%)
Core components (rules, skills, commands, agents, agent memory), session management, built-in and custom slash commands, headless and streaming modes, auto-mode, the CLAUDE.md hierarchy, repository initialisation and settings.json.
Small weight, but if you already use Claude Code daily, these are cheap marks.
Domain 4: Eval, Testing, and Debugging (2.6%)
Identifying error types, choosing a recovery strategy, reading traces to find failure modes, and working out whether a problem sits in your integration code or in the model's output.
Domain 5: Model Selection and Optimization (16.8%)
- LLM fundamentals (5.2%): tokens, context windows, sampling, non-determinism, extended and adaptive thinking, effort levels, and zero-, single- and multi-shot prompting.
- Technical fundamentals (6.1%): working with SDKs that wrap REST APIs, and websockets.
- Model selection and trade-offs (2.7%): Opus versus Sonnet versus Haiku, quality versus latency versus cost, and breaking behaviour changes between model releases.
- Cost and token management (2.8%): token tracking, cost modelling, prompt caching and cache checkpointing.
Domain 6: Prompt and Context Engineering (11.0%)
- Prompt engineering (4.6%): clear instructions, few-shot examples, system versus user placement, output constraints, iterative refinement and input sanitisation.
- Context engineering (3.8%): managing the context window, preventing drift and bloat through tool-output pruning and compaction, and isolating context with subagents.
- Output handling (2.6%): structured output, response validation, defensive parsing, and scepticism toward confident output.
Domain 7: Security and Safety (8.1%)
- AI application security (3.2%): prompt injection, jailbreak defence, untrusted input, data leakage, PII handling, authentication and authorisation.
- Guardrails and safe deployment (2.3%): layered guardrails, least privilege and secure-by-design.
- Identity, secrets and key management (1.6%): handling API keys and credentials across development and production.
- Claude hooks (1.0%): using hooks to block destructive actions.
Domain 8: Tools and MCPs (10.6%)
- Tool implementation (4.4%): function calling, writing tool descriptions, error handling, client-side versus server-side tools, and approval patterns.
- Agentic customisation (4.1%): choosing between built-in tools, custom tools, Skills and MCP servers for a given job.
- MCP server development (2.1%): authoring and deploying servers, MCP resources, tools and prompts, and transports such as stdio.
What the questions look like
Scenario questions where every option sounds plausible, and the right one fits the stated constraint best.
Anthropic's guide includes three sample questions. One example, paraphrased: a team has 10,000 documents to process overnight for a non-urgent report, and cost is the main concern. The right answer is the Message Batches API. Firing synchronous requests in parallel, trimming max_tokens or dropping to the smallest model all miss the actual trade-off, which is batch versus realtime.
The other two samples follow the same shape. For prompt injection in user-submitted web pages, the answer is to treat that content as untrusted, keep it apart from your instructions, and use guardrails or hooks so it cannot trigger sensitive actions. For a reusable internal REST service, the answer is an MCP server rather than pasting data into prompts.
The lesson: read for the constraint (cost, reuse, security, latency) before you read the options. The exam rewards knowing why one approach fits, not just that it exists.
Can you take it? The eligibility catch
Only if you work at a Claude Partner Network organisation and register with that company's email. Personal addresses are rejected, and an individual cannot join the network alone.
This is the step that stops most people, so check it before you buy any prep. If you are on a personal address, read can you take the Claude exams without a company email. If you are eligible, the booking steps are in how to register for Claude certification.
Developer, Architect or Associate?
Take Developer if you write the code; take Architect if you design systems other people build.
Developer – Foundations tests implementation: the API, tools, agents, prompts, configuration. Architect – Foundations is aimed at solution architects designing agent systems. Associate – Foundations suits advisory and project-lead roles, and it is the only Claude exam that does not count toward Partner Network standing. Developer does count.
The full side-by-side is in Claude Certified Architect vs Developer vs Associate, and if you are still deciding whether to sit any of them, see is the Claude Certified Architect worth it.
Five mistakes to avoid
The common thread: preparing for the exam you imagine instead of the one in Anthropic's blueprint.
- Treating it as a Claude Code exam. Claude Code is 3.1%. Applications and Integration is 33.1%. Spend your time accordingly.
- Skipping software engineering basics. REST, JSON, async, version control and refactoring are 7.4% on their own. Strong prompt skills will not cover them.
- Studying from unchecked material. Check any course or practice test against Anthropic's eight official domains and their weights before you rely on it.
- No hands-on build. Anthropic's own guidance is to build and run at least one Claude application that uses the API, integrates tools, and includes basic prompt, context, security and evaluation practices. Scenario questions are much easier when you have hit the problem for real.
- Booking on a feeling. With full-fee retakes and a 14-day minimum wait, guessing is expensive. Book on evidence from practice results by domain.
How to prepare: a plan built on the blueprint
Start with a diagnostic across all eight domains, fix the heaviest weak areas first, build one real app, then book on evidence.
- Read Anthropic's exam guide end to end. Section 6 (the blueprint) is your checklist. Self-assess against every skill.
- Take a diagnostic before any course. Find out which of the eight domains are weak. Ten hours on something you already know is ten hours taken from a gap.
- Prioritise by weight × weakness. A weak Applications and Integration (33.1%) matters far more than a weak Eval domain (2.6%).
- Use Anthropic's free courses for the gaps. On Claude Academy, Building with the Claude API maps to Domains 2, 5 and 6; Introduction to Model Context Protocol and Model Context Protocol: Advanced Topics map to Domain 8; Claude Code in Action, Introduction to agent skills and Introduction to subagents map to Domains 1, 3 and 8. Partner organisations also get certification prep courses in the Anthropic Partner Academy. That mapping is mine, not Anthropic's.
- Build one small app that touches everything. For example: an agent that calls one custom tool and one MCP server, uses prompt caching, validates structured output, treats fetched content as untrusted, and has a hook that blocks a destructive action.
- Practise with scenario questions and review every wrong answer. Turn each mistake into a flashcard and revisit it until it sticks.
- Book when every domain is consistently strong in practice. My rule of thumb is around 80% per domain, held over several sessions. It is a personal heuristic, not a guarantee.
Preparing for Claude Certified Developer – Foundations with StudyTech AI
StudyTech AI has a Claude Certified Developer – Foundations prep path that follows the steps above, starting from a free assessment by domain.
StudyTech AI is a personalised AWS and AI certification prep platform. It finds your knowledge gaps, builds your practice around your skills and gaps, turns your mistakes into flashcards, scores your readiness for every exam domain, and tells you when you are ready to take your exam.
For CCDV-F, that means:
- A free assessment that scores you across the exam domains
- Focused lessons on your weakest domain first
- Wrong answers become flashcards automatically
- Readiness tracked by domain, so you can see when each one is consistently strong
- Personalised mock exams aimed at your gaps
Pricing is simple: $29 a month or $297 a year.
StudyTech AI does not register you for the exam. You still need a Claude Partner Network company email to book through Pearson VUE.
Start a free Claude Developer assessment on StudyTech AI
Verified 7 November 2026 against Anthropic's Claude Certified Developer – Foundations Exam Guide (version 1.0, July 2026), the Anthropic Partner Academy certifications FAQ and Anthropic's certification announcement. Anthropic states the guide may change without notice, so confirm the current version before booking.