CCA-F Exam Prep · First cohort launching soon · Register interest →
Live Weekend Coaching · 100% Online · Zoom
CERTIFICATION PREP — LIVE COACHING

Master the CCA-F exam with live coaching from practitioners who build with Claude

Two-day live intensive covering all 5 exam domains. Add the full prep bundle for 500+ practice questions, domain study guides, exam simulations, and a strategy guide to stand out as a certified Claude architect.

Register Interest

720/1000 is the passing score. We prepare you to score well above it.

Live instruction
500+ practice questions
From ₹25,000
14
Hours of live coaching
500+
Practice questions
5
Domain study guides
3
Exam simulations
Built for engineers at Anthropic Partner Network companies

Prerequisites: This course requires a Claude Pro or Max subscription ($20/month) for interactive teaching prompts, and an Anthropic API key for build exercises. You'll need both active before the bootcamp.

The Claude Certified Architect exam is for builders, not beginners

Anthropic's CCA-F certification tests production architecture decisions. This live coaching is for engineers who want structured preparation with instructor-led guidance to score high on their first attempt.

This is for you if

  • You build production systems using the Claude API, Claude Code, or MCP
  • You have 6+ months hands-on experience with Claude in a professional context
  • You understand agentic loops, tool design, and multi-agent orchestration
  • Your company is part of the Anthropic Partner Network
  • You want live coaching, not just self-study materials

This is not for you if

  • You've only used Claude through the chat interface (claude.ai)
  • You don't have programming experience
  • You're looking for a general AI literacy certification
  • Your company is not in the Anthropic Partner Network
  • You haven't worked with APIs, SDKs, or CLI tools before
!

Anthropic Partner Network requirement: To register for the CCA-F exam, your organization must be part of the Anthropic Partner Network. Current anchor partners include Accenture (~30,000 professionals), Cognizant (~350,000 associates), Deloitte, and Infosys. Check with your company's Anthropic relationship owner to confirm eligibility before beginning your study.

What the Claude Certified Architect exam actually tests

This is not a trivia exam. Every question is a production scenario covering agentic architecture, prompt engineering, MCP integration, and context management. You're the architect — what do you recommend?

60
Scenario-based MCQs
120
Minutes (proctored)
720
Passing score (out of 1000)
Domain Topics Weight
Prompt Engineering & Structured Output Explicit criteria, few-shot prompting, tool_use schemas, validation, batch 20%
Claude Code Configuration & Workflows CLAUDE.md hierarchy, commands, plan mode, CI/CD integration 20%
Tool Design & MCP Integration Tool descriptions, error categories, MCP servers, tool_choice 18%
Agentic Architecture & Orchestration Agentic loop, multi-agent patterns, subagent config, hooks, sessions 27%
Context Management & Reliability Context preservation, escalation, error propagation, provenance 15%
i

The #1 tested concept: Programmatic enforcement vs. prompt-based guidance. When the exam gives you a choice between telling the model to do something (prompt) and making it impossible to do otherwise (code), the code answer is almost always correct. Prompts suggest. Code enforces.

Live coaching includes everything you need to pass

Not just coaching. A complete certification prep bundle — study guides, question banks, build exercises, and exam strategy. One price. No add-ons.

📚

5 Domain Study Guides

In-depth coverage of every exam domain, grounded in three real production scenarios. ~20 hours of structured material for self-study before and after the coaching.

~20 hours of material
🎯

500+ Practice Questions

Scenario-based MCQs matching the real exam format. Every question includes the correct answer, full justification, and an explanation of why each distractor fails.

Anti-patterns, deep dives, mock exams
🛠

Student Handbooks & Build Exercises

Hands-on implementation specs with student repos. Build a multi-agent coordinator, design tool schemas, configure CLAUDE.md hierarchies. Verification checklists included.

~10 hours hands-on work
📈

Practice Exam Simulations

Full 60-question mock exams matching the real CCA-F format. 120 minutes, domain-weighted scoring against the 720/1000 threshold. The closest thing to the real exam.

3 full exam simulations

Exam Strategy Guide

"Always choose" and "always reject" decision rules extracted from exam patterns. Cross-domain concept mapping and 300+ Claude Code feature reference.

Know the patterns before exam day

Structured to build in complexity

Saturday covers foundations. Sunday covers advanced architecture and exam simulation. Each session combines concepts with hands-on practice.

Saturday — Foundations 10 AM – 5 PM IST
10:00

Prompt Engineering & Structured Output

Explicit criteria, few-shot prompting, tool_use schemas, validation-retry loops, fabrication prevention

concepts hands-on
11:30

Break

11:45

Claude Code Configuration & Workflows

CLAUDE.md hierarchy, custom commands, skills, plan mode, path-specific rules, CI/CD integration

concepts build
1:00

Lunch break

2:00

Tool Design & MCP Integration

Tool descriptions as selection mechanism, error categories, tool_choice, build-vs-use decisions for MCP servers

concepts hands-on
3:30

Break

3:45

Day 1 Practice & Review

Domain-specific practice questions, justification walkthrough, gap identification

practice
Sunday — Advanced + Exam Sim 10 AM – 5 PM IST
10:00

Agentic Architecture & Orchestration

Loop lifecycle, multi-agent hub-and-spoke, subagent isolation, hooks, task decomposition, session state

concepts build
11:30

Break

11:45

Context Management & Reliability

Context preservation, escalation triggers, error propagation, scratchpad strategies, context degradation

concepts hands-on
1:00

Lunch break

2:00

Exam Strategy & Cross-Domain Patterns

Always-choose / always-reject rules, cross-domain concept mapping, time management, trap recognition

strategy
3:00

Break

3:15

Full Exam Simulation + Review

60-question timed mock exam, live scoring, gap analysis, personalised study plan for remaining prep

exam sim

The decision rules that pass candidates use

The exam rewards specific architectural instincts. These patterns appear across all 5 domains.

The exam always rewards

  • stop_reason over natural language signals for loop termination
  • Programmatic enforcement (hooks, gates) for financial, security, and compliance
  • Better tool descriptions as the first fix for misrouting
  • Community MCP servers before building custom ones
  • Fresh start + summary injection over stale resumed context
  • Concrete examples over prose descriptions
  • Independent review instances (not self-review)
  • Nullable fields over required fields to prevent fabrication

The exam always penalises

  • Prompt-based solutions for high-stakes scenarios
  • Arbitrary iteration caps as the primary termination mechanism
  • Sentiment-based escalation triggers
  • Batch API for blocking workflows
  • Single-pass reviews of 10+ files
  • Silent error suppression or workflow termination on partial failure
  • Subagents inheriting full coordinator history
  • Self-reported confidence as an escalation signal

This is what the exam looks like

Every practice question follows this format: a production scenario, four plausible answers, and a detailed breakdown of why one is correct.

Domain — Agentic Architecture
A multi-agent research system has a coordinator that spawns three specialist subagents (market analysis, competitive intelligence, technical assessment). The coordinator passes its full conversation history to each subagent for context. Reports are coming back with duplicated findings and the system is running 3x slower than expected.
What is the most likely root cause?
  • A. The subagents need higher token limits to process the full history
  • B. Subagents should receive only their specific task context, not the coordinator's full history
  • C. The coordinator needs a deduplication step after collecting reports
  • D. The system should run subagents sequentially instead of in parallel
Why B is correct: Subagents in a hub-and-spoke pattern should be isolated. Passing the coordinator's full conversation history causes attention dilution — subagents process irrelevant context, leading to duplicated findings and slower execution. The fix is at the input, not the output.

Why C is tempting but wrong: Deduplication treats the symptom (duplicated findings) but not the cause (context leakage). The 3x slowdown would persist.

One weekend. Exam-ready Monday.

The coaching teaches concepts in 2 days. The full prep bundle gives you the system to drill them over the next 2 weeks.

LIVE COACHING
Live Coaching
First cohort — coming soon
Sat & Sun · 10 AM – 5 PM IST
₹25,000 incl. GST
~$300 USD · GST invoice available
  • 2-day live coaching (14 hours)
  • Session recordings
  • Student handbooks
  • Community access
  • 1 domain study guide
  • 100+ practice questions
  • 1 exam simulation
  • Exam strategy guide + reference
Register Interest

Not satisfied after Day 1? Full refund.

Register your interest. We'll notify you when the first cohort opens.

I'm interested in
Live Coaching ₹25,000
Coaching + Full Prep ₹35,000
Optional

Priority access for early registrants. No payment required now.

You're on the list. We'll be in touch when registration opens.

If you fail the exam, you can't retake for 6 months. ₹10,000 more for the full prep is insurance against that.

Common questions

Saturday covers foundations — prompt engineering, Claude Code configuration, and MCP integration. Sunday covers agentic architecture, context management, exam strategy, and ends with a full timed exam simulation. Each session combines concept walkthroughs with hands-on practice. Live on Zoom, 10 AM – 5 PM IST both days.
You can take the coaching regardless of your organisation. However, to register for the actual CCA-F exam, your company must be part of the Anthropic Partner Network. Current anchor partners include Accenture, Cognizant, Deloitte, Infosys, and TCS. Check with your company's Anthropic relationship owner to confirm eligibility.
There are two tiers. Live Coaching (₹25,000) includes 2 days of live instruction, session recordings, student handbooks, ~50 practice questions from the sessions, and community access. Live Coaching + Full Prep (₹35,000) adds 5 domain study guides (~20 hours), 500+ practice questions with full justifications, 3 exam simulations, exam strategy guide, and a cross-domain reference library. The full prep bundle is what you use for the 2 weeks of self-study after the coaching weekend.
The coaching covers all 5 domains in depth. Most participants need 15–20 additional hours of self-study using the included study guides and practice questions. We recommend spreading this over 1–2 weeks after the coaching. The exam simulation on Day 2 gives you a personalised gap analysis so you know exactly where to focus.
Study guides are Markdown files (readable in any editor). Practice questions include full justifications in the same format. Build exercises come as student repos with specs and verification checklists. Interactive teaching prompts run inside Claude Code with a Pro or Max subscription. No videos, no LMS, no login.
No. This is an independent exam preparation resource built by The Crux AI. It is not affiliated with, endorsed by, or officially associated with Anthropic. "Claude" and "Claude Code" are trademarks of Anthropic, PBC.
Yes — we provide a GST invoice on request, suitable for L&D reimbursement. Many companies in the Anthropic Partner Network have AI upskilling budgets that cover exactly this. Email us at hello@thecrux.ai for a proforma invoice before purchase if needed.
If you're not satisfied after Day 1, email us for a full refund — no questions asked. hello@thecrux.ai

The certification window is open. Get a head start.

The Claude Certified Architect - Foundations is Anthropic's first professional certification. Early movers will have the credential before it becomes table stakes. If you fail, you can't retake for 6 months.

Register Interest

Priority access when the first cohort opens.

Live instruction
500+ questions
GST invoice available