Exam Overview
Version 1.0 · Effective July 2026 · Exam code: CCAR-F
CCAR-F is the entry point to Anthropic’s Architect track. It tests whether you can design a production Claude system rather than prompt one: how an agentic loop terminates, when a subagent earns its context, what belongs in an MCP server, how a CLAUDE.md shapes Claude Code’s behaviour, and what to do when a context window fills mid-task.
Below is the exam on paper: every domain, weight and rule, taken from Anthropic’s guide. Everything needed to pass it is on this site, free and without a sign-up.
Who the exam is for
Section titled “Who the exam is for”Solution architects and technical practitioners designing Claude-powered agentic systems. Anthropic sets no prerequisite exam and no formal experience bar, and the credential is awarded on exam performance alone. That doesn’t make it an entry-level paper. The questions assume you’ve built something with Claude and hit the problems the blueprint asks about, which is why every lesson here ends in a build exercise rather than a summary.
Exam facts at a glance
Section titled “Exam facts at a glance”| Item | Detail |
|---|---|
| Exam code | CCAR-F |
| Number of items | 60 |
| Item format | Multiple-choice and multiple-response; each item states how many responses to select |
| Time limit | 120 minutes |
| Delivery | Through Pearson VUE, online proctored or at a test centre |
| Passing score | Scaled score of 720 on a scale of 100–1,000 |
| Anthropic’s exam fee | $125 USD per attempt |
| Validity | 12 months from the date the credential is awarded |
| Result reporting | Pass/fail with scaled score, plus percent-correct by domain on the score report |
| Retakes | waiting periods of 14, 30, and 90 days after failed attempts, up to four attempts in a rolling twelve months, fee per attempt |
| Renewal | free on-time renewal via a non-proctored assessment on the Anthropic Partner Academy; a lapsed credential requires the full exam at the full fee |
All figures from Anthropic’s Exam Guide v1.0 (July 2026). The exam fee is Anthropic’s charge for the official exam, not a price of this site; nothing here requires payment today.
Exam blueprint
Section titled “Exam blueprint”Domain weights are the guide’s own figures. Every task statement below is a written lesson on this site, so the blueprint doubles as the table of contents.
Domain 1: Agentic Architecture & Orchestration
Section titled “Domain 1: Agentic Architecture & Orchestration”27%
- 1.1 Agentic Loops
- 1.2 Multi-Agent Orchestration
- 1.3 Subagent Invocation and Context Passing
- 1.4 Workflow Enforcement and Handoff
- 1.5 Agent SDK Hooks
- 1.6 Task Decomposition Strategies
- 1.7 Session State and Resumption
18%
- 2.1 Tool Interface Design
- 2.2 Structured Error Responses
- 2.3 Tool Distribution & Tool Choice
- 2.4 MCP Server Integration
- 2.5 Built-in Tools
Domain 3: Claude Code Configuration & Workflows
Section titled “Domain 3: Claude Code Configuration & Workflows”20%
- 3.1 CLAUDE.md Hierarchy, Scoping, and Modular Organisation
- 3.2 Custom Slash Commands and Skills
- 3.3 Path-Specific Rules for Conditional Convention Loading
- 3.4 Plan Mode vs Direct Execution
- 3.5 Iterative Refinement Techniques
- 3.6 CI/CD Integration
Domain 4: Prompt Engineering & Structured Output
Section titled “Domain 4: Prompt Engineering & Structured Output”20%
- 4.1 System Prompts with Explicit Criteria
- 4.2 Few-Shot Prompting
- 4.3 Structured Output with Tool Use
- 4.4 Validation, Retry, and Feedback Loops
- 4.5 Batch Processing Strategies
- 4.6 Multi-Instance and Multi-Pass Review
Domain 5: Context Management & Reliability
Section titled “Domain 5: Context Management & Reliability”15%
- 5.1 Context Window Management
- 5.2 Escalation & Ambiguity Resolution
- 5.3 Error Propagation in Multi-Agent Systems
- 5.4 Codebase Exploration & Context Degradation
- 5.5 Human Review & Confidence Calibration
- 5.6 Information Provenance & Multi-Source Synthesis
How to prepare for it here
Section titled “How to prepare for it here”Take the diagnostic first. Finding out which of the 5 domains you can move through quickly beats starting at domain 1 by default. Then work the curriculum in blueprint order and sit the timed mock once the heavier domains are behind you. 30 lessons and 250+ practice questions, all free, no sign-up.
Take the diagnostic · Browse the curriculum · Quick reference
Frequently asked questions
Section titled “Frequently asked questions”What does the CCAR-F exam cover?
60 items in 120 minutes across 5 domains: Agentic Architecture & Orchestration (27%), Tool Design & MCP Integration (18%), Claude Code Configuration & Workflows (20%), Prompt Engineering & Structured Output (20%), Context Management & Reliability (15%). Scoring is scaled from 100 to 1,000, and you need 720 to pass.
Who is the CCAR-F exam for?
Solution architects and technical practitioners designing Claude-powered agentic systems. That’s Anthropic’s stated audience, and the full profile sits in section 3 of the official exam guide. There’s no prerequisite exam and no formal experience requirement; the credential is awarded on exam performance alone.
Is the preparation on this site really free?
Yes, and there’s no sign-up. All 30 lessons, 250+ practice questions, the quick-reference cards, the glossary and the full timed mock exam are open to everyone. The only money involved anywhere is Anthropic’s own $125 exam fee, which is paid to Anthropic when you sit the real thing.
How long does preparation take?
It depends on how much production Claude work you’ve already done. The curriculum is 30 lessons weighted to match the blueprint, so the 27% domain carries more of them than the 15% one. Most people work through a domain at a time, using the diagnostic first to find out which ones they can move quickly through.
How do I register for the official exam?
Registration runs through Anthropic’s Partner Academy, which is currently open to people at organisations in the Claude Partner Network and needs a company email on that domain. That access gap is exactly why this site exists: complete preparation for people who can’t get through the partner door.
What happens if I fail?
Waiting periods of 14, 30, and 90 days after failed attempts, up to four attempts in a rolling twelve months, fee per attempt. Your score report breaks the result down by domain, so you can see which areas cost you the attempt before booking another.
Is this site affiliated with Anthropic?
No. This is an independent community resource, not affiliated with, endorsed by, or sponsored by Anthropic. Every exam fact on this page comes from Anthropic’s published exam guide and is attributed to it.
An independent community resource — not affiliated with or endorsed by Anthropic. Request access to the official exam.