About the CCA-F exam
What you need to know before you sit the Claude Certified Architect Foundations exam.
Anthropic's official study guide covers 23% of what's actually tested.
The CCA-F Trainer covers the other 77% — including 8 topics the study guide doesn't mention at all.
“Get very familiar with the Claude Certified Architect – Foundations Certification Exam Guide. It contains all the things you need to know about the exam, such as the scenarios that will appear on the exam, the domains and skill areas, preparation exercise ideas, and the types of questions you’ll encounter on the Practice and Certification Exams. This step is very important to your success. Unless you are already very familiar and have practice with Claude Code, Agent SDK, the Claude API, and Model Context Protocol, skipping this step will likely result in you not passing the Certification Exam.”
— Anthropic Education FAQ, “What are the basics?”
Anthropic's own FAQ is direct: the Exam Guide alone won't carry you through unless you already have hands-on experience with every technology it covers. That's the gap this app is built to close.
60 questions, 120 minutes. Delivered online via ProctorFree. Government-issued photo ID required at session start. No notes permitted; no AI assistance during the exam.
“All questions on the exam are multiple choice format. Each question has one correct response and three incorrect responses (distractors). Select the single response that best completes the statement or answers the question. Distractors are response options that a candidate with incomplete knowledge or experience might choose. Unanswered questions are scored as incorrect; there is no penalty for guessing.”
— Anthropic Skilljar exam materials
Subjects tested: Claude Code, Claude Agent SDK, Claude API, and Model Context Protocol (MCP) — the core technologies used to build production-grade applications with Claude.
“The exam has a pass or fail designation. The exam is scored against a minimum standard established by subject matter experts. Your results are reported as a scaled score of 100–1,000. The minimum passing score is 720. … Please note that results are not immediately available after taking the exam. Your score will be emailed within 7 business days.”
— Anthropic Skilljar exam materials
No penalty for guessing — always select an answer. Scaled scoring helps equate scores across exam forms with slightly different difficulty levels.
One attempt per registration. Verbatim Anthropic Skilljar FAQ. There is no option to retake without purchasing a new registration at $99 standard (or $47.50 with a partner promo code). Your sitting is your one shot.
“Please do not attempt this exam if you are not ready. If you fail this exam today, you will not be able to re-take it for 6 months.”
— Anthropic Skilljar exam materials (Welcome & what to expect lesson; repeated verbatim on the proctoring consent page)
Practical implication: one attempt per registration plus a 6-month wait if you don't pass. One sitting, six months on the line — make it count.
$99 USD per registration. Purchase through Anthropic's Skilljar portal.
Some Anthropic partners provide a promo code at enrollment — $47.50 (50% off) at checkout for eligible accounts. The official Anthropic FAQ states: “Some participants may have a promo code through their partner organization.” The benefit is a promo code applied at purchase, not a fee waiver.
Topic-by-topic comparison of the official study guide against the exam's assessed content:
The official guide covers 23% of what's tested at sufficient depth. The remaining 77% — 30+ concepts — require knowledge from hands-on experience, the Anthropic docs, or this app. Concepts like why the same Claude session that generated your code is biased when reviewing it, how --output-format json works in CI pipelines, and why personal skills silently override project skills with the same name.
Scenario-based. Questions are drawn exclusively from the 8 official scenarios listed below. Each exam form presents 4 of the 8 official scenarios — candidates see 4 scenarios per sitting, drawn from the same 8-scenario inventory. Drilling all eight removes the possibility of being caught off-guard by the draw.
These are the only scenarios from which questions are drawn:
- Customer Support Resolution Agent
- Code Generation with Claude Code
- Multi-Agent Research System
- Developer Productivity with Claude
- Claude Code for CI/CD
- Structured Data Extraction
Questions will not test knowledge of:
- Fine-tuning Claude models or training custom models
- Claude API authentication, billing, or account management
- Detailed implementation of specific programming languages or frameworks (beyond what's needed for tool and schema configuration)
- Deploying or hosting MCP servers (infrastructure, networking, container orchestration)
- Claude's internal architecture, training process, or model weights
- Constitutional AI, RLHF, or safety training methodologies
- Embedding models or vector database implementation details
- Computer use (browser automation, desktop interaction)
- Vision/image analysis capabilities
- Streaming API implementation or server-sent events
- Rate limiting, quotas, or API pricing calculations
- OAuth, API key rotation, or authentication protocol details
- Specific cloud provider configurations (AWS, GCP, Azure)
- Performance benchmarking or model comparison metrics
- Prompt caching implementation details (beyond knowing it exists)
- Token counting algorithms or tokenization specifics
60 questions, scenario-bundled, calibrated harder than the official practice exam. Use your Full Sim score as your readiness signal. Target ≥820 on the simulated exam, which maps to Anthropic's recommended 900+ on their official practice exam.
Take only when you are consistently scoring above 720 on Full Sim. Community feedback indicates the practice exam runs easier than the real exam. Scoring well on it before your Full Sim is ready can produce false confidence.
Anthropic's official recommendation: score above 900/1000 on the practice exam before attempting the real one.
Per Anthropic Exam Policy §6 (effective April 1, 2026): certifications must be renewed before the Certification Term expires. If the term lapses without renewal, you must re-earn the certification from scratch — renewal is not possible after expiry. The policy defines the Term but does not specify its length in the publicly available documentation.
Beta-cert risk: CCA-F is currently offered in a partner-exclusive beta state. Per Exam Policy §5: “If you passed a practice or beta Exam or an Exam related to a beta product and earned a Certification as a result, your Certification may expire once the final Exam or production product goes live.” Badge holders may need to re-earn their certification when the production version ships.
“This exam is proctored by ProctorFree, a third-party service that helps us verify who is taking the exam and keep the certification credible for everyone who earns it. … When you set up proctoring, ProctorFree scans your face to confirm you are the person taking the exam. This biometric data is used to verify your identity, prevent fraud, and maintain security. It is not used for any other purpose and is deleted within 30 days. ProctorFree also records your exam session so it can flag potential misconduct. That recording may be shared with Anthropic and, if relevant, the organization through which you’re taking the exam.”
— Anthropic Skilljar exam materials (proctoring page)
- Requirements: webcam, screen sharing, stable internet connection
- Biometric face scan used for identity verification
- Biometric data deleted within 30 days
- Session recording may be shared with Anthropic and your partner organization
- Alternative verification available — contact academy-support@anthropic.com
Three rules enforced by ProctorFree throughout your session:
- No AI assistance — using AI products or services during the exam is prohibited.
- Government-issued photo ID — must match the name on your registration.
- Audiovisual monitoring — ProctorFree records audio and video for the duration of your session.
Violations can result in score invalidation, suspension from future exams, or permanent prohibition — with no fee refund.
Appeal policy: Misconduct or enforcement decisions can be appealed within 14 calendar days by emailing academy-support@anthropic.com with “appeal” in the subject line. The 14-day window applies to enforcement decisions, not to retake timing.
Accommodations: Reasonable accommodations for disabilities are available — contact academy-support@anthropic.com with “accommodation” in the subject line before your exam date.
Geographic availability: Anthropic reserves the right to exclude specific regions and countries from the program at any time.
The Trainer does not predict your real-exam outcome. Readiness ≥720 on Full Simulated Exams is a necessary but not sufficient condition. The exam tests judgment under realistic constraints; this app trains for that, but real-exam variance is real.
Among the people publicly sharing their scores, results cluster in the upper band — roughly 890 to 985. Selection bias applies: people who barely pass don't post screenshots. Don't use that range as a calibration target. Use your Full Sim score.
Ready to cover the 77%?