Exam Format
Remember
- 60 questions. 120 minutes. Pass ≈ 720 / 1000.
- Every question is a scenario. No pure theory questions.
- You get 4 of 6 scenario blocks. Prepare all 6.
Key numbers (from the official exam guide, v1.0 July 2026)
| Item | Value |
|---|---|
| Exam code | CCAR-F |
| Questions | 60. Mix of multiple-choice and multi-response (each states how many to pick; partial credit is community-reported, not in the guide) |
| Time | 120 minutes |
| Typical finish time | 70–105 minutes (community reports, not in the guide) |
| Passing score | 720 on a 100–1000 scale |
| Fee | $125 USD |
| Certificate validity | 12 months |
| Platform | Pearson VUE (OnVUE online, or test center) |
| Score report | Pass/fail + percent-correct per domain |
Official domains and weights
| # | Domain | Weight |
|---|---|---|
| 1 | Agentic Architecture & Orchestration | 27% |
| 2 | Tool Design & MCP Integration | 18% |
| 3 | Claude Code Configuration & Workflows | 20% |
| 4 | Prompt Engineering & Structured Output | 20% |
| 5 | Context Management & Reliability | 15% |
Domain 1 is the biggest → patterns/patterns.mdx and scenarios/multi-agent-research.mdx are your highest-yield reads.
The 6 scenario blocks
The exam randomly picks 4 of these (numbered in the official guide's order):
| # | Scenario (official title) | Difficulty (as reported by test-takers) |
|---|---|---|
| 1 | Customer Support Resolution Agent | Hard — tricky escalation judgment |
| 2 | Code Generation with Claude Code | Medium |
| 3 | Multi-Agent Research System | Medium |
| 4 | Developer Productivity with Claude (Agent SDK) | Medium |
| 5 | Claude Code for Continuous Integration | Hard — specific flags and setup |
| 6 | Structured Data Extraction | Medium |
Each block has its own file in scenarios/.
Heads-up: one test-taker reported extra scenarios beyond these 6 (Agent Skills, long-document processing, conversational AI). The official v1.0 guide is explicit that the bank is exactly 6 — so that report is unverified and may reflect a different exam or a newer form. Either way, the underlying concepts are covered by patterns/ and the fact folders (Agent Skills content is part of Domain 3).
What it feels like (from people who passed)
- Real exam is harder than every mock (CertSafari, Udemy, Skilljar).
- Mock score 95–100% ≈ real score ~80%.
- Usually 2 options are easy to kill. The last 2 are both plausible.
- Questions are long. Reading stamina matters as much as knowledge.
Next: One-Page Cheat Sheet