Home › Study guides › CCAR-P › Domain 2

CCAR-P · Domain 2 of 7 · 5 lessons · about 111 min

Domain 2: Claude Models, Prompting & Context Engineering

Choosing a Claude model, system prompts and guardrails, prompting techniques, context and token use, and reuse with caching and Skills. 13% of CCAR-P.

This domain is 13% of CCAR-P, about 8 of the 63 questions. It covers what the model is and what it sees: which Claude model a workload runs on, how its instructions are written, and how the limited space of its context is spent.

The vocabulary: the context window is everything the model reads for one request, measured in tokens (pieces of words), and it has a fixed size. A system prompt sets the model's role and rules for a whole conversation. Few-shot prompting shows examples; chain-of-thought asks the model to reason before answering. Prompt caching lets a repeated opening of a request be reused instead of processed again; a Skill packages instructions and files that Claude loads only when a task needs them.

The recurring test: match the model and the prompt to the task, and spend context on what the task needs. The official sample question has the pattern: static content goes first and is cached, rather than cut down or moved somewhere it cannot be reused. Options that pick the biggest or the smallest model regardless of fit, or throw away required context to save tokens, are usually the distractors.

What the exam guide tests

The official CCAR-P guide lists 5 objectives for this domain. Exam questions are written against them, and so are the lessons: each row says where it is covered.

Lessons

1. 2.1 Selecting Claude models: trade-offs, constraints and the evidence to decideHow an architect picks a Claude model per workload: the tiers compared, platform, residency and lifecycle limits, and evals on your own data that decide.22 min
2. 2.2 System prompts, templates and prompt guardrailsWhat belongs in the system prompt and what in the user turn, how to version and test a prompt template, and the guardrails a prompt can and cannot hold.23 min
3. 2.3 Zero-shot, few-shot and chain-of-thought: choosing the technique per decisionWhen clear instructions are enough, when examples must teach your conventions, when reasoning earns its tokens, and how an eval set settles the choice.22 min
4. 2.4 Optimising the context window: select, compact and measure tokensWhy a bigger window is not the fix, how to choose what enters each request, keep long sessions lean with compaction and clearing, and measure tokens.21 min
5. 2.5 Prompt reuse: caching, modular prompts and Agent SkillsHow to cache a repeated prompt prefix, keep shared rules in owned and versioned modules, and package know-how as Skills that load only when needed.23 min

Practice question

From the CCAR-P bank, tagged to this domain. Every answer option is explained. Nothing is stored, nothing to sign up for.