# CCAR-F Practice Exam 2

A second full-length, blueprint-weighted practice exam with new, harder, multi-constraint items — the booking gate after Exam 1.

import PracticeExam from '../../../components/PracticeExam.astro';
import { Steps } from '@prosefly/astro-components';

This is a **second, tougher** full-length sitting: 60 brand-new items, none repeated from Practice Exam 1, the domain pages, or the scenarios page. Compared with Exam 1, Exam 2 leans harder into the way the real Architect exam actually reads — **multi-constraint stems** (a speed goal *and* a compliance rule; a cost target *and* a quality bar), and more items qualified with **FIRST**, **MOST cost-effective**, **BEST**, and **TWO**. Several stems layer two anti-patterns as distractors so eliminating one is not enough.

Every item is tagged to a **domain (D1–D5)** and one of the **six scenarios (S1–S6)**, exactly as the blueprint tests them. Treat this as your go/no-go gate: sit Exam 1 first to learn the format, then use Exam 2 to decide whether to book.

## Domain distribution (matches the blueprint)

| Domain | Weight | Items |
| --- | --- | --- |
| D1 · Agentic Architecture and Orchestration | 27% | 16 |
| D2 · Claude Code Configuration and Workflows | 20% | 12 |
| D3 · Prompt Engineering and Structured Output | 20% | 12 |
| D4 · Tool Design and MCP Integration | 18% | 11 |
| D5 · Context Management and Reliability | 15% | 9 |

## Scenario coverage

Items are drawn across all six reference scenarios so you cannot pass by mastering only one cluster:

| Scenario | Focus |
| --- | --- |
| S1 · Customer Support Resolution Agent | Loop termination, escalation, hooks, idempotency, tool security |
| S2 · Code Generation with Claude Code | CLAUDE.md/settings precedence, hooks, plan mode, subagents |
| S3 · Multi-Agent Research System | Orchestrator-workers, explicit context, partial failure, context editing |
| S4 · Developer Productivity | Server-side tools, MCP scopes, direct-call vs model tool |
| S5 · Claude Code for CI/CD | Headless JSON, least privilege, multi-pass review, Batch API |
| S6 · Structured Data Extraction | Structured outputs, Fable 5.1 tool-choice, per-type metrics, caching arithmetic |

## How to use both exams

<Steps>

1. **Sit Practice Exam 1 first**, untimed if you like, to internalise the question anatomy and the ten anti-patterns.

2. **Study your weak domains** on the domain pages and the anti-patterns page before returning.

3. **Sit Practice Exam 2 timed (120 min)** as the *booking gate*. Aim for **≥ 80% raw** (≥ 48/60) with no single domain below ~70%.

4. **Compare per-domain results** across both exams. A domain that is weak on *both* is your real gap — drill it before you book.

</Steps>

:::tip[Booking signal]
The real exam is scaled 100–1000 with a pass at **720**, reported per domain. Because Exam 2 is harder, a consistent **≥ 80% raw here**, with D1 (27%) solid, is a strong go signal. If D1 or the anti-patterns are shaky, hold off.
:::

## Score interpretation

| Raw score (of 60) | Approx. band | Reading |
| --- | --- | --- |
| 54–60 (90–100%) | Well above pass | Exam-ready across all domains |
| 48–53 (80–88%) | Above pass | Ready; shore up any single weak domain |
| 43–47 (72–78%) | Around the line | Borderline; drill weakest domain before booking |
| 36–42 (60–70%) | Below pass | Not ready; revisit D1 and the anti-patterns |
| < 36 (< 60%) | Well below | Restudy the domain pages before re-attempting |

## Take the practice exam

The **interactive mode** runs a timed sitting one question at a time and ends with your score, a per-domain breakdown and a full correction; the **review mode** underneath lists every question with its options one per line and the answer hidden until you ask for it.

<PracticeExam exam="ccar-f" set={2} />