Pass CCA-F
in six months.
Free to start.
Implementation-first prep for the Anthropic Claude Certified Architect Foundations exam. Take the 10-Q diagnostic to find your weakest domain, work the 25 concepts, then prove it on the 30-Q mock. Server-tracked mastery, cohort leaderboard, no fluff.
What is the Claude Architect (CCA-F) exam?
Anthropic's foundational architecture certification. 60 questions across five domains, 720 of 1000 to pass, 90-minute clock. Practical, judgment-heavy, no API trivia.
The Anthropic Claude Certified Architect Foundations exam (CCA-F) is the industry credential for builders who design with Claude as a primary reasoning system. It tests architectural judgment, not API trivia, across five domains, with a real cut score and a strict 90-minute clock.
- Questions
- 60
- Pass mark
- 720/1000
- Duration
- 90 min
- Domains
- 5
- D1Agentic loops + tool design27%
- D2Context engineering18%
- D3Production scenarios20%
- D4Evaluation + measurement20%
- D5Safety + responsible AI15%
The 30-question mock on this site pulls by these exact weights, so your readiness signal mirrors the real exam, not a flat distribution.
How do I pass the CCA-F exam?
Three surfaces, in order. Take the diagnostic first; it routes you to the concepts you actually need - depth on MCP, agentic loops, the CLAUDE.md hierarchy, agentic tool design, and multi-agent research patterns. The mock proves it. The adaptive plan keeps you sustainable.
Take the 10-Q diagnostic.
Stratified across the 5 CCA-F domains. Get your weakest-domain signal and a study route. No sign-up required.
Start diagnosticProve it on the 30-Q mock.
Real blueprint weights, timed, 720 to pass. Signed-in users get three lifetime attempts (server-enforced).
Open mockFollow your 4-week plan.
Built post-mock from your specific misconceptions. Concept tiles highlight what's in your plan; mastery is server-tracked.
See curriculumIs Claude Architect prep really free?
The short answer is yes. The longer answer is that every gate on the path to passing (diagnostic, mock, mastery tracking, adaptive plan, cohort) is at zero cost.
Yes. Every gate on the path to passing is at $0.
There is no paid tier today. Sign-up unlocks the full stack below; the diagnostic is open to anonymous visitors so you can validate the platform before deciding. The credential is real. The prep doesn't cost you anything.
Every “mastered” concept is gated on server-tracked signals: dwell time, correct practice, completed mocks. Other prep sites mark things complete when you click. Here you earn it, and the score on your cohort leaderboard reflects that.
Why does this certification matter now?
The Indian IT sector is in the middle of its biggest skill rebalance in twenty years. The people who can prove they understand Claude as a system primitive are getting pulled forward. The numbers and signals are public.
The Indian IT sector is hiring for this exact skill, right now.
The shift from labor arbitrage to AI-architecture is the biggest skill rebalance the sector has seen in twenty years. Claude is the new system primitive; the people who can prove they understand it are the ones getting pulled forward.
- Anthropic and OpenAI now ship managed service lines - the same Indian IT majors who built on Java + cloud labor arbitrage are hiring CCA-F-grade architects to lead AI-first delivery.
- TCS, Infosys, Wipro, Cognizant, and Hexaware have all opened AI architecture practices in the last 18 months. The bottleneck is qualified people, not budgets.
- Recruiters increasingly screen for certificate-anchored skills: provable concept mastery on the blueprint, not just years of experience on a CV.
Where are 472 students right now?
Every signed-in student maps to one of six mascot stages, driven by a 0–1000 readiness score. The journey is the same whether you're student 12 or student 12,000.
Frequently asked questions.
The most common questions about the CCA-F exam and this prep platform. Answers are written so AI Overviews, Perplexity, and ChatGPT can lift them, and so you don't have to read the docs to make a decision.
The Claude Certified Architect Foundations exam (CCA-F) is Anthropic's foundational credential for builders who design with Claude as the primary reasoning system. The 60-question exam covers five domains: agentic loops and tool design (27%), context engineering (18%), production scenarios (20%), evaluation and measurement (20%), and safety and responsible AI (15%). Passing score is 720 out of 1000, and the time limit is 90 minutes.
Most candidates pass within 4 to 6 months at a sustainable pace of roughly 4 to 6 hours per week. The adaptive plan on this platform re-paces from your diagnostic result, so a candidate strong in agentic loops but weak in evaluation will spend less time on the former and more on the latter. There is no required minimum study time. The platform reflects how much real, server-tracked practice you've done, not how many hours you've claimed.
720 out of 1000. The 30-question mock on this platform uses the same 720 cut score so your readiness signal mirrors the real exam, not an inflated friendly number. Score thresholds for the readiness composite (0–1000) are calibrated to predict the live exam outcome within ±5%.
Yes. The diagnostic, 30-question mock, concept mastery tracking, 4-week adaptive plan, cohort leaderboard, mascot evolution, and badges are all $0. There is no paid tier today. The diagnostic stays open for anonymous visitors so you can validate the platform before signing up; the full stack unlocks with a free account.
Anthropic publishes the exam blueprint and reference documentation. This platform turns the blueprint into a server-tracked practice loop. Every "mastered" concept is gated on dwell time plus correct practice questions plus completed mocks, not on a click. Other prep sites mark items done when you check a box; here you have to earn it, which is why the cohort leaderboard means something.
No. The 10-question diagnostic is open to anonymous visitors and runs in the browser. Sign-up is only required to take the full 30-question mock, get a personalized 4-week adaptive plan, and join a cohort leaderboard. Anonymous diagnostic results reset on tab close.
Thirty questions pulled across the five CCA-F domains by real blueprint weight: D1 (27%, ~8 questions), D2 (18%, ~5–6), D3 (20%, ~6), D4 (20%, ~6), D5 (15%, ~4–5). The mock is timed, server-scored, and 720 out of 1000 is the pass line. Signed-in accounts get three lifetime attempts, enforced server-side. Question explanations are unlocked after submission.
The adaptive plan rebuilds around your specific wrong-answer misconceptions, not a generic retake syllabus. If you missed three D4 evaluation questions, the plan front-loads D4 concept reads, evaluation scenarios, and targeted practice drills before letting you re-mock. Anthropic's retake window is typically 30 days; the 4-week plan is sized to that.
A cohort is 5–25 peers grouped automatically by signup month and tier. Each Monday at 00:00 UTC, the platform takes a snapshot and re-ranks every member. Your readiness score on the leaderboard is a 0–1000 composite: 50% concept mastery (blueprint-weighted), 30% best mock score in the last 90 days, 10% scenario walkthroughs, and 10% daily-streak score.
No. Telemetry stays on-platform and is used only to drive your adaptive plan and the cohort snapshot. Cohort handles are pseudonymous in V1 (e.g., "Loop A4F"); privacy controls on /account let you toggle visibility. No third-party advertising trackers run on the practice surface. The full privacy story is on the methodology page.