Home › Certifications › CCDV-F Certification Guide
Anthropic Certification Guide · 2026CCDV-F Certification Guide 2026: Claude Certified Developer – Foundations
The complete guide to Anthropic’s developer credential — all eight domains with their exact decimal weightings, why the blueprint is ten times heavier at one end than the other, what it costs, the salary hike the skills command in India and the US, and a 6-weekend roadmap to pass it.
Authorized Training & Technology Partners
Claude
Claude Certified Developer – Foundations (CCDV-F)
The credential this guide coversClaude Certified Developer – Foundations (CCDV-F) is Anthropic’s engineering credential for building on Claude rather than with it. It is a 120-minute proctored exam of 53 questions via Pearson VUE, it costs US$125, you pass at a scaled 720 out of 1000, and it covers eight domains. You write code: Messages API, streaming, caching, batching, agents, tool schemas, MCP servers, guardrails and evals.
One fact should reshape your entire study plan before you read anything else. The CCDV-F blueprint is brutally uneven. Applications and Integration is 33.1% of the exam. Claude Code is 3.1%. That is more than a tenfold difference — and most candidates spend roughly equal time on both.
CCDV-F exam at a glance
| Attribute | Detail |
|---|---|
| Credential | Claude Certified Developer – Foundations (CCDV-F) |
| Issued by | Anthropic |
| Level | Developer — the engineering tier, above Associate and below Architect |
| Format | 53 questions, multiple-choice and multiple-response |
| Duration | 120 minutes, proctored |
| Scoring | Scaled; 720 out of 1000 to pass |
| Domains | Eight, weighted from 33.1% down to 2.6% |
| Exam fee | US$125, paid directly to the exam provider at registration |
| Delivery | Pearson VUE — online proctored or at a test centre |
| Blueprint version | v1.0, effective July 2026 |
| Validity | 12 months from the date awarded |
| Coding required | Yes — Python or TypeScript, plus REST, JSON and async fundamentals |
What is the Claude Certified Developer credential?
CCDV-F certifies that you can build and operate production software on the Claude API. Where CCAO-F asks whether you can use Claude well at work, CCDV-F asks whether you can ship an application that calls it correctly, cheaply, safely and observably.
It sits one rung below CCAR-F, and the split between them is clean: CCDV-F is about implementation — the API surface, integration patterns, cost engineering, tool schemas, MCP servers. CCAR-F is about architecture — whether the system you designed survives contact with production.
CCDV-F domains and weightings
Eight domains, published to one decimal place, summing to exactly 100%. The spread is the story: the top two domains are half the exam between them, while the bottom two together are under 6%.
CCDV-F exam blueprint — share of exam by domain
Blueprint v1.0, effective July 2026. Bars scaled to the largest domain.
Source: CCDV-F public exam blueprint v1.0, effective July 2026. Weightings sum to 100.0%. Domain 5 covers the LLM and software-engineering fundamentals the exam assumes — tokens, context windows, sampling and non-determinism, plus REST, JSON, async, version control and SDLC.
Applications & Integration (33.1%) is worth more than ten times Claude Code (3.1%). Add Fundamentals (16.8%) and you are at 49.9% — almost exactly half the exam — from two domains. If your study time is not weighted like this, you are optimising for marks that do not exist. Claude Code and Evals together are 5.7%; treat them as a single evening, not a weekend.
The eight domains as flashcards
Ordered by exam weight, not by syllabus order. Study top-down.
Applications & Integration
The Messages API: requests, roles, multi-turn state. Streaming, vision and extended thinking. Prompt caching mechanics and cache checkpointing. The Message Batches API — 24-hour windows, custom_id correlation, and when batch is the wrong choice. Token tracking and cost modelling. Invoking Claude through third-party vendors.
Fundamentals
LLM mechanics the exam assumes you know: tokens, context windows, sampling, non-determinism. Plus the software-engineering foundations it tests directly — REST, JSON, async, version control and SDLC. Configuration management: CLAUDE.md, settings.json, model pinning and prompt versioning.
16.8% of examAgents & the Claude Agent SDK
Workflow versus agent — the decision criteria. The agentic loop: tool_use versus end_turn, keeping tool results in history. The Claude Agent SDK, custom loops and harnesses, self-hosted versus Anthropic-hosted. Subagents and manager hierarchies. Hooks for deterministic actions. Frameworks in scope: LangGraph, Strands, PydanticAI.
Prompt & Context Engineering
Context window management: pruning, compaction, isolation via subagents. Instruction clarity, few-shot design, system versus user placement, output constraints. Input sanitisation and iterative refinement. Structured output patterns, response validation and defensive parsing.
11.0% of examTools, Function Calling & MCP
Tool use and function calling: schema design and description writing that makes models select correctly. Client-side versus server-side tools, approval patterns, error handling. MCP server development — tools, resources, prompts, stdio versus sockets. The trade-off matrix: built-in tools versus custom tools versus Skills versus MCPs.
10.6% of examSecurity, Safety & Guardrails
Prompt injection awareness and mitigation, jailbreak defence, untrusted input handling. Data leakage prevention and PII handling. Guardrails through hooks to block destructive actions. Secrets, credentials and API key management across dev and prod, with least privilege.
8.1% of examClaude Code
Rules, Skills, Commands, Agents and Agent Memory. CLAUDE.md hierarchy and repository initialisation. Session management, slash commands, headless and streaming modes. settings.json configuration. Small domain, easy marks — do not overinvest.
3.1% · smallest but oneEvals & Debugging
Error type identification and recovery strategies. Trace analysis: isolating integration-layer failures from model-output failures. The smallest domain on the exam at 2.6%, but the skill you will use most often once you are actually shipping.
2.6% · smallest domainIs CCDV-F worth it?
Where it genuinely helps
- It is the first Claude credential that evidences engineering ability. Associate proves fluency; Developer proves you can ship. That is the rung where the compensation conversation actually changes.
- Cost engineering is a rare, checkable skill. Prompt caching and batching are a third of the biggest domain, and being able to cut an LLM bill by half is one of the few AI skills a finance team can verify.
- MCP is still early. Being able to author and deploy your own MCP server is a differentiator today in a way it will not be in two years.
- Claude Partner Network firms. Several with large Indian delivery operations have direct commercial reasons to value certified developers, since partner tier and staffing credibility depend on it.
Where it will not carry you
- It assumes you can already code. Python or TypeScript, plus REST, JSON and async. There is a whole domain on software-engineering fundamentals and it is not optional.
- It is not an architecture credential. If you are designing multi-agent systems and judging whether they survive production, that is CCAR-F.
- It expires in 12 months. A recurring commitment, not a trophy.
What the CCDV-F skill set actually pays
The honest framing first. No certificate has a salary attached to it, and there is no credible published figure for a Claude-certification-specific uplift. What is documented is what the skills pay — and CCDV-F is the rung where those skills become engineering skills.
What the hike looks like in rupees
The reported agentic premium applied at its midpoint (~40%) to published Indian software engineering bands.
Entry · 0–2 yrs
Mid · 3–5 yrs
Senior · 6–9 yrs
Lead · 10+ yrs
Claude and LLM engineering pay by market
India and the US shown as two separate charts, because rupee and dollar bands are different measures on different scales — a shared axis would misrepresent both.
India — annual CTC
Claude / LLM application engineer roles, ₹ lakh per annum
Bar length maps the upper bound of each band against a ₹52 L scale. Sources: Glassdoor India AI Architect/Engineer data (July 2026, avg ~₹35 L, typical range ~₹20.75–47.75 L); reported ₹20–40 L+ for engineers with RAG, agentic workflow and system-design skills.
United States — annual base
Claude / LLM engineering roles, US$ thousands
Bar length maps each figure against a $200K scale. Sources: ZipRecruiter Claude AI job postings ($57K–$160K, May 2026); published prompt-engineering ranges of $60K–$250K+ in the US.
Generalist vs Claude application engineer
Generalist software engineer vs Claude / agentic application engineer
India, ₹ lakh per annum — modelled from the reported 30–50% agentic premium
Method: generalist bands drawn from published Indian software engineering ranges; the specialist bar applies the reported 30–50% agentic premium at its midpoint (~40%). Illustrative modelling, not survey data — treat it as direction, not a quotation.
Compensation figures are compiled from independent, publicly available industry sources and are shown for role context. They are not a guarantee of pay in any specific market, company or outcome, and 360DT does not promise a salary result from any certification or programme.
Why the blueprint is weighted this way
1. Integration is where real projects actually fail
Applications & Integration is a third of the exam because that is where production breaks: mis-handled streaming, no caching, batch used when realtime was needed, token costs nobody modelled. Anthropic weighted the exam toward the failures it sees, not toward the features that demo well.
2. Cost engineering became a first-class skill
Prompt caching, cache checkpointing and the Message Batches API sit inside the biggest domain for a reason. As LLM spend moved from experiment budget to line item, the ability to halve a bill without degrading output became something engineering leaders hire for specifically.
3. Claude Code is deliberately small here
At 3.1%, Claude Code is almost a rounding error on CCDV-F — because it is weighted heavily on the Architect exam instead, at 20%. The family is designed so the tiers do not duplicate each other. Knowing that saves you a weekend.
Your 6-weekend CCDV-F roadmap
Twelve sessions, weighted to the real blueprint, ending in two full timed mocks. Every module ships something.
Claude API deep dive
Messages API: requests, roles, multi-turn state. Streaming, vision and extended thinking. Prompt caching mechanics and when they pay off. SDK setup in Python or TypeScript. LLM fundamentals: tokens, context windows, sampling, non-determinism. Covers Domains 1 & 2 — 49.9% combined. Ship: a multi-modal Claude app skeleton.
Integration patterns and cost engineering
Message Batches API: 24-hour windows, custom_id correlation, and when batch is wrong. Caching plus checkpointing. Token tracking and cost modelling. REST, JSON, async, version control and SDLC. Ship: cut a real app’s cost by roughly half, documented.
Agents and the Claude Agent SDK
Workflow versus agent. The agentic loop: tool_use versus end_turn, tool results in history. Custom loops and harnesses, self-hosted versus Anthropic-hosted. Subagents, manager hierarchies, hooks. LangGraph, Strands and PydanticAI. Covers Domain 3 — 14.7%. Ship: a tool-using support agent.
Tools, function calling and MCP servers
Schema design and description writing that makes models select correctly. Client-side versus server-side tools, approval patterns, error handling. MCP server development — tools, resources, prompts, stdio versus sockets. Covers Domain 5 — 10.6%. Ship: your own MCP server integrating a real service.
Claude Code, prompt and context engineering
One focused session on Claude Code’s 3.1% — Rules, Skills, Commands, Agents, Memory, CLAUDE.md hierarchy, settings.json. Then context management: pruning, compaction, subagent isolation, structured output and defensive parsing. Covers Domains 4 & 7 — 14.1% combined.
Security, evals and the exam sprint
Attack your own agent then fix it: injection defence, jailbreak handling, PII, guardrail hooks, secrets and least privilege. Trace analysis separating integration failures from model-output failures. Two full timed 53-question mocks with domain-wise reports, then Pearson VUE registration. Covers Domains 6 & 8 — 10.7%.
How to prepare for CCDV-F
- Spend half your time on two domains. Applications & Integration plus Fundamentals is 49.9%. Everything else competes for the other half.
- Actually implement caching and batching. Reading about prompt caching teaches you nothing about cache checkpointing or when batch is the wrong call — and both are examined.
- Write one MCP server end to end. The tool-design domain is answerable from experience and guessable from theory; the difference shows in your score.
- Cap Claude Code and Evals at one evening. Together they are 5.7% of the paper. Time spent beyond that is time taken from the 33.1% domain.
- Take 53-question timed mocks, reviewed by domain. An evenly-weighted mock badly misrepresents this exam — more than most, because the real spread is 10×.
- Work through a domain-weighted video course first if you are starting cold. Our breakdown of the Claude Certified Developer (CCDV-F) course by Siddharth Sharma maps all 11 sections and 38 lectures to these eight domains, so you can see exactly which weighting each section serves before you spend the time.
Learn from Anthropic Authorized Instructors
360DT delivers Claude certification prep live, taught by Anthropic Authorized Instructors alongside Microsoft, AWS, NVIDIA and Cisco certified trainers — our full international faculty.
Vikas Mittal
Google and Anthropic Authorized AI Instructor
Veteran technology trainer with more than 25 years of rich industry experience spanning Google, Microsoft, and emerging AI platforms. An exceptional mentor recognized for proven teaching excellence.
Deep
Anthropic Authorized Instructor and AI Cloud Specialist
Seasoned technology leader with over 20 years of experience across AI, cloud computing, and enterprise training. Known for simplifying complex ideas and delivering practical, real world learning.
Sid
Anthropic Authorized Instructor and Azure Data Engineer
Dynamic corporate trainer and consultant with more than 10 years of experience across Microsoft Fabric, Azure Data Engineering, and modern cloud platforms. Focused on clear, practical learning that sticks.
Bipeen
AWS, Microsoft, and Anthropic Certified Trainer
AI, machine learning, and cloud transformation expert with more than 25 years of experience. A renowned global corporate trainer and conference speaker known for practical, outcome driven sessions.
Shantanu Pandey
Microsoft, Google, NVIDIA and HPE Certified Trainer
Director of Engineering at Meteoros Automation, ranked in Microsoft's Top 100 Trainers of 2025, holding MCT, Google Cloud, NVIDIA and PeopleCert credentials, with 500+ trainings delivered since 2011.
Arshad Ahmad
Microsoft Certified Trainer (MCT), Cybersecurity & Power Platform Expert
Technology Trainer with 15 years in strategic client delivery, holding MCT, Azure Solutions Architect Expert, Cybersecurity Architect Expert and Power BI Data Analyst credentials, with 200+ trainings delivered.
Sravia
NVIDIA Authorized Instructor and AI Data Specialist
Passionate AI and data evangelist with over 10 years of experience across machine learning, MLOps, and cloud native technologies. Dedicated to delivering impactful, engaging enterprise training programs.
Johan
Microsoft Business Applications MVP and Microsoft Certified Trainer
D365 Customer Service Lead – Europe at Avanade, and founder of the Power Platform School. A Microsoft Business Applications MVP and Microsoft Certified Trainer based in London, specializing in Dynamics 365 Customer Service and Power Platform.
Akim Nyamande
Microsoft, CompTIA, Juniper and Cisco Certified Trainer
IT training facilitator with over five years of experience and certifications across Microsoft, CompTIA, Juniper, and Cisco. Previously a Network Administrator before moving into technical training, now delivering hands-on courses in networking, systems administration, and cybersecurity fundamentals.
Ali El Khatib
Cisco Certified Systems Instructor (CCSI) and Microsoft Certified Trainer
Infrastructure Engineer at RHUH with 13 years in the training field, holding CCSI, MCSE, CCNP Routing and Switching, CCNP Security, CompTIA and Microsoft Azure certifications.
Florian Garcia Compte
Cisco Certified Systems Instructor (CCSI)
Partner Director at AVAANZA FORMACION, a Cisco Learning Partner based in Madrid. Certified Cisco Systems Instructor (CCSI No. 21053) specializing in Cisco Data Center (ACI/SDN), Nexus, and Wireless, delivering official Cisco certification courses including CCNA, CCNP Enterprise, and CCNP Data Center.
Six weekends, weighted exactly like the real blueprint
The 360DT Claude Certified Developer Prep Course teaches all eight domains in their real proportion and tells you the weight of every module up front, so nothing you study is wasted. You leave with a production Claude application, an agent on the Agent SDK, and your own deployable MCP server — on your GitHub, defensible in an interview.
Explore the CCDV-F Prep Course
CCDV-F frequently asked questions
What is the biggest domain on the CCDV-F exam?
Applications & Integration at 33.1% — a third of the exam in a single domain, covering the Messages API, streaming, vision, prompt caching, the Message Batches API and cost modelling. Add Fundamentals at 16.8% and two domains account for 49.9%, almost exactly half the paper. The smallest domain, Evals & Debugging, is 2.6%.
How much does the CCDV-F exam cost?
US$125, paid directly to the exam provider at registration — the same as the Architect-level CCAR-F, and more than the US$99 Associate exam. Training fees are separate from the exam fee.
Do I need to code for CCDV-F?
Yes. CCDV-F assumes Python or TypeScript plus REST, JSON and async fundamentals, and there is an entire 16.8% domain covering LLM and software-engineering foundations. If you do not code, start with CCAO-F instead, which is explicitly non-technical.
What is the CCDV-F passing score and format?
53 questions in 120 minutes, and you pass at a scaled 720 out of 1000. It is proctored through Pearson VUE, either online or at a test centre. Note the question count differs from the Associate and Architect exams, which are both 60 items.
CCDV-F or CCAR-F — which should I take?
CCDV-F if you are implementing: API integration, cost engineering, tool schemas, MCP servers. CCAR-F if you are designing: agentic architecture, orchestration and context reliability at system level. The tiers deliberately avoid overlap — Claude Code is 3.1% on CCDV-F but 20% on CCAR-F.
How much salary hike can CCDV-F give me?
The certificate itself carries no salary, and no credible published figure attaches an uplift to a Claude certification specifically. What is documented is that agentic AI engineering commands a reported 30–50% premium over generalist software engineering, widening to 2–3× for senior multi-agent talent. Applied to Indian bands at the midpoint that is roughly +₹2.5 L at entry, +₹5 L at mid, +₹9 L at senior and +₹13 L at lead. Modelled figures for direction, not a quotation.
How long does it take to prepare for CCDV-F?
Roughly six weekends for a working developer who already ships in Python or TypeScript — the twelve-session roadmap on this page. Add time if you have never implemented prompt caching, batching or a tool-calling loop, since those sit inside the heaviest domain.
How long is the CCDV-F credential valid?
12 months from the date it is awarded, in line with the rest of the Claude credential family. Plan for re-certification as a recurring commitment.
Sources and further reading
- Anthropic — Claude Certified Developer – Foundations credential, delivered via Pearson VUE
- CCDV-F public exam blueprint v1.0, effective July 2026 — eight domains: Applications & Integration 33.1%, Fundamentals 16.8%, Agents & Agent SDK 14.7%, Prompt & Context Engineering 11.0%, Tools, Function Calling & MCP 10.6%, Security, Safety & Guardrails 8.1%, Claude Code 3.1%, Evals & Debugging 2.6%
- Published CCDV-F exam specification — 53 questions, 120 minutes, scaled 720/1000 to pass, US$125, 12-month validity
- ZipRecruiter (May 2026) — Claude AI job postings, $57K–$160K
- Glassdoor India (July 2026) — AI Architect/Engineer average ~₹35 L, typical range ~₹20.75–47.75 L
- Independent 2026 industry reporting on the agentic AI engineering premium (30–50% over generalist software engineering, widening to 2–3× at senior multi-agent level) and on the absence of a credible Claude-certification-specific salary figure
360DT is an independent training provider. Certification exams are administered independently by Anthropic through Pearson VUE and are not included in programme tuition. Exam details are accurate as of 20 August 2026; always confirm current format, pricing and domain weightings with Anthropic before booking.