Check you can register before you study
The exam guide says no course or credential is required. That is a statement about content, not about access: registration is restricted to Claude Partner Network organizations and rejects personal email addresses, and that restriction is published nowhere in the guide. Both halves are below, because either one alone is misleading.
Who may register
Not in the exam guideCertification is open to people at Claude Partner Network organizations, and registration rejects personal email addresses. This is not stated in the exam guide.
“Yes. Certification is available to people at Claude Partner Network organizations. Registration requires a partner email address on a recognized company domain — personal email addresses will not work.”
Partner Academy FAQPrerequisites
In the exam guideNo course or prior credential is required, and Professional does not require Foundations. This is about content, not eligibility — see who may register.
“There are no mandatory prerequisites or courses required to sit this exam. The experience above is recommended, not required. The credential is awarded based on exam performance alone.”
Exam guideAccess inside the network
Not in the exam guideThere is no second gate once your organization is in the network — every partner can register. The gate is membership itself, not seniority or an invitation.
“Any individual at a partner organization in the Claude Partner Network. There are no access gates: every CPN partner can register”
Partner Academy FAQFoundations before Professional
Not in the exam guideYou may sit Professional without holding Foundations, and Foundations never upgrades into it.
“Foundations is the natural place to start and there's no formal prerequisite, so you can take Professional without holding Foundations. Foundations does not convert or upgrade to Professional automatically.”
Partner Academy FAQWho delivers the exam
Not in the exam guidePearson VUE administers the exams, and a test-center appointment is managed through your own Pearson account.
“Test center appointments can be rescheduled or cancelled up to 24 hours prior to your scheduled appointment through your Pearson account.”
Pearson VUE
Quotations are verbatim from the linked source. Links last confirmed to resolve on 2026-08-24.
What sitting it involves
Exam code CCDV-F, written against guide v1.0 (July 2026).
About 136 seconds an item if you spend it evenly.
A scaled score on a 100–1000 range. It is not 72% of items correct.
Charged in full on every attempt, including retakes.
On-time renewal is free and unproctored. A lapsed credential means the whole exam again, at full fee.
Online-proctored or at a test center.
How the result is decided, and what happens after it
How the exam is scored
In the exam guideYou are measured against a fixed standard, not against other candidates. 720 is a standard-set cut score on a 100–1,000 scale — it is not 72% correct.
“a criterion-referenced assessment: each candidate is measured against a fixed performance standard, not against other candidates. You pass by demonstrating the knowledge and skills defined in the blueprint, not by outperforming a percentage of peers.”
Exam guidePer-domain percentages
In the exam guideYour score report shows a percentage per domain, but no domain has its own pass line. Only the total scaled score decides the result.
“Section-level percentages are provided to help you understand your performance and are not used to determine your pass or fail result, which is based on your total scaled score.”
Exam guideRetakes
In the exam guideWaits lengthen with each failure — 14, then 30, then 90 days — capped at four attempts per exam per rolling year. Each attempt is charged in full, and the limit is per exam.
“Waiting periods increase with each failed attempt: 14 days after the first, 30 days after the second, and 90 days after the third. You may take an exam up to four times within a rolling twelve-month period. Limits apply per exam, so not passing one exam does not prevent you from registering for a different one. The exam fee applies to each attempt.”
Exam guideRenewal after 12 months
In the exam guideOn-time renewal is a free, non-proctored assessment. Let the credential lapse and you pay the full fee and sit the whole exam again — which is what makes the 12-month expiry a diary entry rather than a recurring cost.
“To renew on time, you review what has changed since you certified and complete a free, non-proctored assessment on the Anthropic Partner Academy. There is no fee for on-time renewal. If your credential lapses, you must retake the full exam at the full fee to regain certified status.”
Exam guideAppeals
In the exam guideFourteen days to appeal, through Pearson VUE. Neither the cut score nor an individual item can be appealed.
“You may appeal a decision within 14 days of the date you are notified of it, or, for a concern about your exam result, within 14 days of your exam date. … The standard-setting outcome and the content of individual exam items are not subject to appeal.”
Exam guide
What each sub-domain is actually worth
The CCDV-F guide publishes a weight for all 25 sub-domains, not just the 8 domains — so the expected number of items each one contributes to a 53-item form is derivable rather than guessed. Every figure below is computed from the published weights.
Where the study time should go
12 of the 25 sub-domains are expected to be worth under two items each, and Identity, Secrets, and Key Management (0.85) and Claude Hooks (0.53) may not appear on your form at all. Prep material for this exam leads on agents, MCP servers and Claude Code.
Meanwhile Claude Application Design, Software Engineering Foundations, Claude API Mechanics — three sub-domains of ordinary application engineering — are worth about 12.1 items between them, roughly 23% of the whole exam. Applications and Integration alone is 33.1%.
| Domain and sub-domain | Weight | Expected items | Band |
|---|---|---|---|
| 2Applications and Integration | 33.1% | 17.5 | 6 sub-domains |
| Claude Application Design | 8.6% | 4.6 | 2+ items |
| Software Engineering Foundations | 7.4% | 3.9 | 2+ items |
| Claude API Mechanics | 6.8% | 3.6 | 2+ items |
| Configuration Management | 4.1% | 2.2 | 2+ items |
| Understanding Requirements | 3.4% | 1.8 | 1–2 items |
| Systems Life Cycle | 2.8% | 1.5 | 1–2 items |
| 5Model Selection and Optimization | 16.8% | 8.9 | 4 sub-domains |
| Technical Fundamentals | 6.1% | 3.2 | 2+ items |
| LLM Fundamentals | 5.2% | 2.8 | 2+ items |
| Cost and Token Management | 2.8% | 1.5 | 1–2 items |
| Model Selection and Tradeoffs | 2.7% | 1.4 | 1–2 items |
| 1Agents and Workflows | 14.7% | 7.8 | 3 sub-domains |
| Agent Construction with Claude | 5.3% | 2.8 | 2+ items |
| Agent Patterns and Frameworks | 4.9% | 2.6 | 2+ items |
| Agent Architecture | 4.5% | 2.4 | 2+ items |
| 6Prompt and Context Engineering | 11% | 5.8 | 3 sub-domains |
| Prompt Engineering | 4.6% | 2.4 | 2+ items |
| Context Engineering | 3.8% | 2.0 | 2+ items |
| Output Handling | 2.6% | 1.4 | 1–2 items |
| 8Tools and MCPs | 10.6% | 5.6 | 3 sub-domains |
| Tool Implementation | 4.4% | 2.3 | 2+ items |
| Agentic Customization | 4.1% | 2.2 | 2+ items |
| MCP Server Development | 2.1% | 1.1 | 1–2 items |
| 7Security and Safety | 8.1% | 4.3 | 4 sub-domains |
| AI Application Security | 3.2% | 1.7 | 1–2 items |
| Guardrails and Safe Deployment | 2.3% | 1.2 | 1–2 items |
| Identity, Secrets, and Key Management | 1.6% | 0.85 | ≤1 item — may not appear at all |
| Claude Hooks | 1% | 0.53 | ≤1 item — may not appear at all |
| 3Claude Code | 3.1% | 1.6 | 1 sub-domain |
| Claude Code Operation | 3.1% | 1.6 | 1–2 items |
| 4Eval, Testing, and Debugging | 2.6% | 1.4 | 1 sub-domain |
| Debugging and Error Handling | 2.6% | 1.4 | 1–2 items |
These are expected values, not a form. The guide calls its weights “the approximate proportion of scored items”, so a sub-domain worth 0.53 items means “usually zero, sometimes one” — and a sub-domain worth half an item can still be the hardest thing you are asked.
The denominator is the published 53. The guide states 53 items and weights them against “scored items”, without saying whether all of them are scored or some are unscored pretest items. If some are unscored, every figure here is proportionally generous. We do not know, and we would rather say so than pick a quieter number.