Can you register today?
The CCD-F has no prerequisite courses, exams, or prior experience requirements - the credential is awarded on exam performance alone. Registration for these certifications is through the Anthropic Partner Network (on the Anthropic Partner Academy) - partner-network access is required; you then schedule your proctored session with Pearson VUE. Walk this path top to bottom before you schedule.
- No prerequisite courses or exams. You do not need to pass another Claude certification or complete a course first - no Claude certification is a prerequisite for another. The CCD-F credential is awarded on exam performance alone; the recommended experience helps you prepare but is never checked at registration.
- Register through the Anthropic Partner Network. Registration for the CCD-F is handled on the Anthropic Partner Academy and requires Anthropic Partner Network access - you register through an Anthropic partner, not by public self-signup. Review the exam guide and the Certification Terms and Conditions, then complete checkout. The CCD-F lists at $125; the checkout price reflects any partner-tier discount.
- Create a Pearson VUE account and schedule. Follow the confirmation instructions to sign in to Pearson VUE, then choose an available date and either online proctoring or a test center. You can cancel or reschedule up to 24 hours before your appointment; changes made within 24 hours forfeit the fee.
- Sit the exam. 53 questions, 120 minutes, delivered proctored (online or at a Pearson VUE test center). Present a valid, unexpired, government-issued photo ID that matches your registration name exactly.
Your CCD-F credential is valid for 12 months from the award date. Renew on time with a free, non-proctored assessment on the Anthropic Partner Academy - no fee. If it lapses, you must retake the full exam at the full fee.
Wait 14 days after your 1st fail, 30 days after the 2nd, 90 days after the 3rd. You can attempt the exam up to 4 times within a rolling 12-month period; the exam fee applies to every attempt.
What gets tested, weighted by importance
The CCD-F samples each domain at the official blueprint weight. Spend study time proportionally.
| Code | Domain | Weight | Key concepts |
|---|---|---|---|
| CCDVF-D1 | Agents & Workflows Choosing between a workflow and an agent (and the decision criteria that drive it), constructing agents with the Claude Agent SDK, custom agent loops and harnesses, self-hosted vs. Anthropic-hosted deployment, and applying common agent design patterns - tool-use loops, sub-agents, memory, context-window management - via frameworks such as Strands, LangGraph, or PydanticAI. | 14.7% | |
| CCDVF-D2 | Applications & Integration By far the heaviest domain - about a third of the exam. Translating business/functional requirements and systems-lifecycle concepts into working code, Claude API mechanics (messages, tools, streaming, vision, thinking, caching, third-party invocation, Batches API), core software-engineering practice (REST APIs, JSON, async programming, version control, code review), Claude application design across interfaces (Claude Code, Desktop, claude.ai, API, SDKs), and configuration management (CLAUDE.md, settings.json, model version pinning, plugin dependencies). | 33.1% | |
| CCDVF-D3 | Claude Code A narrow but hands-on domain: Claude Code's core components (Rules, Skills, Commands, Agents, Agent Memory), its interaction modes (session management, built-in and custom slash commands, headless mode, streaming mode, auto-mode), the CLAUDE.md hierarchy, repository initialization, and settings.json configuration. Note this is only 3.1% here - it is CCA-F, the Architect exam, that weights the Claude Code CLI heavily (a dedicated 20% domain). | 3.1% | - |
| CCDVF-D4 | Evaluation, Testing & Debugging The lightest domain on the exam: debugging and error-handling technique for Claude applications - identifying error types, selecting a recovery strategy, tracing failures to isolate whether the problem sits in the integration layer or in model output. | 2.6% | |
| CCDVF-D5 | Model Selection & Optimization LLM fundamentals (tokens, context windows, sampling, non-determinism), the technical fundamentals of wrapping REST APIs and websockets, choosing between Opus, Sonnet, and Haiku on quality-latency-cost trade-offs and adaptive-thinking support, and managing tokens and cost through budgeting, caching, and cache checkpointing. | 16.8% | |
| CCDVF-D6 | Prompt & Context Engineering Context and memory management (window management, preventing context drift and bloat via tool-output pruning/compaction, context isolation through subagents), prompt-engineering principles (instruction clarity, few-shot placement, iterative refinement, input sanitization), and output handling - structured-output patterns, response validation, defensive parsing, and skepticism toward confident-sounding output. | 11% | - |
| CCDVF-D7 | Security & Safety AI application security (prompt-injection awareness and mitigation, untrusted-input handling, data-leakage prevention, PII handling), guardrails and secure-by-design deployment (content-policy layering, least-privilege identity and access management), Claude Hooks for deterministic guardrails against destructive actions, and identity/secrets/API-key management. | 8.1% | |
| CCDVF-D8 | Tools & MCPs Tool implementation practice (function calling, tool description writing, error handling, client-side vs. server-side tool patterns), MCP server development (authoring, deployment, resources, prompts, stdio vs. socket transports), and choosing among built-in tools, custom tools, Skills, and MCPs for a given use case. | 10.6% | - |
Where to spend your study hours: Applications & Integration and Model Selection & Optimization together are about 50% of the exam. Note Claude Code is only 3.1% here - it is CCA-F, the Architect exam, that weights the Claude Code CLI heavily (a dedicated 20% domain), not the Developer exam.
The 8 CCD-F Competency Domains and Their Exam Weights
The same blueprint summarized as discrete Q&A so AI assistants can lift one domain at a time. Weights mirror the table above.
- What is CCDVF-D2 (Applications & Integration) and how much of the CCD-F does it weight?
- Applications & Integration is by far the heaviest domain at 33.1% of the CCD-F - about a third of the exam. It covers translating requirements into code, Claude API mechanics (messages, tools, streaming, vision, caching), software-engineering foundations, application design across interfaces, and configuration management.
- What is CCDVF-D5 (Model Selection & Optimization) and what weight does it carry?
- Model Selection & Optimization is 16.8% of the CCD-F, the second-heaviest domain. It tests LLM fundamentals (tokens, context windows, sampling), technical fundamentals of wrapping REST APIs and websockets, choosing between Opus/Sonnet/Haiku, and managing tokens and cost through caching.
- What is CCDVF-D1 (Agents & Workflows) and what weight does it carry?
- Agents & Workflows is 14.7% of the CCD-F. It covers the workflow-vs-agent decision, constructing agents with the Claude Agent SDK, custom agent loops, and agent design patterns (tool-use loops, sub-agents, memory) via frameworks like LangGraph or Strands.
- Why is Claude Code only 3.1% of the CCD-F exam?
- Claude Code (CCDVF-D3) is a narrow, hands-on domain at just 3.1% - it tests core components (Rules, Skills, Commands, Agents), interaction modes, and configuration, but not at the depth the CCA-F Architect exam tests it. Note the contrast: it is CCA-F, not the Developer exam, that dedicates a full 20% domain to Claude Code configuration and workflows.
- What is the lightest-weighted domain on the CCD-F exam?
- Eval, Testing, and Debugging (CCDVF-D4) at just 2.6% - the smallest domain across any of the 4 Claude certifications. It still counts: it covers debugging and error-handling technique, identifying error types, and tracing failures to isolate whether a problem sits in the integration layer or the model output.
- What is the full domain-weight breakdown for the CCD-F exam?
- The eight competency domains sum to 100%: Agents and Workflows 14.7%, Applications and Integration 33.1%, Claude Code 3.1%, Eval/Testing and Debugging 2.6%, Model Selection and Optimization 16.8%, Prompt and Context Engineering 11.0%, Security and Safety 8.1%, Tools and MCPs 10.6%. These are non-round percentages taken directly from the version 1.0 exam guide's blueprint table.
Applied CCD-F scenarios
CCD-F currently ships 6 dedicated scenarios: Production Claude Chat Application, Streaming and Tool-Use Integration, Custom MCP Tool Server, Model Selection and Cost Routing, Agent Evaluation and Testing Harness, and Prompt Injection Defense Layer - each walked as a full build (components, data flow, code-level decisions, failure modes). Filter to CCD-F on the scenarios hub to study them in isolation.
Browse CCD-F scenarios →Environment rules catch more candidates than the questions do
The format is tightly defined in Anthropic's version 1.0 exam guide. Read the proctoring rules before you schedule - a clean workspace and a full room sweep are non-negotiable, whether you test online or at a Pearson VUE test center.
- Items53 scored questions
- Time limit120 minutes total
- Item formatMultiple-choice and multiple-response
- Passing scoreScaled score of 720 on a 100-1,000 scale
- Result reportingPass/fail with scaled score, plus percent-correct by domain
- DeliveryOnline-proctored or test center via Pearson VUE
- IdentificationValid, unexpired, government-issued photo ID
- Workspace rulesEffectively closed and proctored
- Prohibited itemsMobile phones, smart watches, headphones, study materials, recording devices
- MisconductInvalidation of result, revocation of credential, ban from future exams
- AppealsWithin 14 days, submitted to Pearson VUE
- Cancel / rescheduleUp to 24 hours before your appointment
Ready to get CCD-F certified?
CCD-F has no prerequisite courses or exams, but registration is through the Anthropic Partner Network (on the Partner Academy) - register there, then schedule your proctored session with Pearson VUE. List price is $125; the checkout price shown reflects any discount that applies to your partner tier.
Free practice, live today
A free diagnostic, a weakest-domain warm-up, and a full-length timed 53-question mock are all live now for CCD-F, sampled from its question bank across every domain above.
Who should take CCD-F?
AI/ML engineers & senior SWEs shipping production Claude apps, agents, and workflows. If that description matches your current role and you have hands-on exposure to Claude in a production setting, CCD-F is the right certification to target - the 8 domains above (and their weights) are the exact shape of what the exam expects you to already know.
How CCD-F differs from the other Claude certs
CCD-F is the hands-on developer track: its heaviest domain, Applications & Integration, is weighted far above anything in CCA-F (the flagship). Where CCA-F (the flagship) tests broad agentic-architecture judgment and CCA-P tests senior architecture and governance, CCD-F is aimed squarely at engineers shipping Claude-powered code day to day.
19 questions answered
Every Q is phrased as a real Google search query and tagged by intent.