ANTHROPIC-CCDV-F: Claude Certified Developer - Foundations (CCDV-F)
Browse certifications
Exam Resources
Official learning paths, exam details, skills measured, and community resources to supplement your study.
About the CCDV-F Exam
Before buying: Anthropic currently limits exam registration to employees of Claude Partner Network organisations, using a recognised company email domain. Buying this independent revision bank does not establish eligibility or include an exam voucher. Check Anthropic’s current certification eligibility FAQ before planning an exam.
Claude Certified Developer - Foundations covers the implementation of applications built with Claude: the API, agent workflows, integration, developer tooling, model choices, context, security, and MCP.
This bank contains 250 original single-answer revision questions mapped to the public CCDV-F Exam Guide, version 1.0, July 2026. It is not official Anthropic exam material or endorsed training. The real exam also includes multiple-response items. Our smaller domains are deliberately oversampled for study breadth; this is not a proportional exam simulation, and a practice score does not predict an exam result.
Who Should Take This Exam?
The guide describes developers who build and integrate applications using Claude. No mandatory training or skill prerequisite is listed, but the partner-only registration restriction still applies. A developer should be comfortable reading application code, following request and response lifecycles, and reasoning about errors, state, permissions, and tests.
This Developer bank focuses on implementation decisions. It is not a substitute for practising with the actual tools, nor a claim that every product feature covered here will appear on an exam.
Exam Quick Facts
These figures come from the July 2026 guide. Check the official guide and current FAQ before booking; the practice-bank price is separate from the exam fee.
| Detail | Value |
|---|---|
| Credential | Claude Certified Developer - Foundations (CCDV-F) |
| Questions | 53 multiple-choice and multiple-response items; each item states how many responses to select |
| Time | 120 minutes |
| Pass Score | Scaled score of 720 on a 100-1,000 scale |
| Exam Fee | USD $125 |
| Delivery | Proctored online and/or at a test centre, per program policy |
| Validity | 12 months after award |
| Eligibility | Claude Partner Network employees with a recognised employer email domain |
| Official Guide | Anthropic’s July 2026 CCDV-F exam guide |
Exam Domains & Weights
The official decimal weights total 100%. Practice counts are an editorial study allocation, not the exam’s distribution. We provide at least 24 questions per domain, including the two small developer-tooling and debugging domains.
| Domain | Official Weight | Practice Questions |
|---|---|---|
| Agents and Workflows | 14.7% | 33 |
| Applications and Integration | 33.1% | 64 |
| Claude Code | 3.1% | 24 |
| Eval, Testing, and Debugging | 2.6% | 24 |
| Model Selection and Optimization | 16.8% | 33 |
| Prompt and Context Engineering | 11.0% | 24 |
| Security and Safety | 8.1% | 24 |
| Tools and MCPs | 10.6% | 24 |
| Total | 100% | 250 |
Practice Questions - 250 Original Scenarios
Try 20 questions free, with no account needed. Full bank access costs US$9 for one year, with explanations, reasoning for the wrong answers, public-source references, and progress tracking.
Every practice item has four options and one correct answer. The real exam’s multiple-response format is not reproduced by this bank. Use the questions to identify topics to practise, not to estimate an official pass mark.
Study Tips
- Start with Applications and Integration, the largest official domain at 33.1%. Trace requests, responses, streaming events, application state, and configuration through a small working application.
- Separate a workflow from an agent. Identify which steps the application controls and where the model chooses its next action. Inspect the actual tool-call and result messages.
- Debug with evidence. Capture request identifiers, error categories, and relevant trace events. Distinguish a retryable service error from a request that must be corrected.
- Measure model and context choices. Compare representative outputs and token use against a defined task, rather than assuming a larger model or more context always improves the result.
- Keep permissions outside untrusted text. Practise safe tool boundaries, credential handling, and the differences between instructions, retrieved material, and authorised actions.
- Use the current first-party documentation. Claude Code, the API, Agent SDK, and MCP features change. Read the linked source when a question raises a product-specific detail.