CCDV-F Prompt and Context Engineering Practice Question
You are integrating Claude 3.5 Sonnet into a customer support application that maintains long, multi-turn conversations. After about 40 turns, you notice Claude begins contradicting policy details it stated earlier in the same session, even though the policy text is still included in the system prompt. The conversation history is passed in full on each request. What is the most effective structural change to preserve instruction adherence across the session?
⚠ Common exam trap
The trap here is assuming that a longer context window automatically means the model will keep honoring instructions placed anywhere in that context.
Answer choices
Why each option matters
Answer the question above first, then reveal the full breakdown to understand why each option is right or wrong.
Correct answer & explanation
✓
Restate the critical policy instructions in the system prompt and periodically re-inject a condensed reminder of them in recent turns.
Instruction adherence degrades in long conversations because relevant constraints become diluted across many tokens. Keeping durable rules in the system prompt maintains their authority, and re-injecting condensed reminders near the current turn restores salience so the model honors them. Together these preserve consistency without abandoning the conversational history the application relies on.
Answer analysis
Option-by-option breakdown
For each option: why learners choose it and why it is or isn't the right answer here.
- ✗
Increase the max_tokens parameter so Claude has more room to generate responses that include the policy details.
Why it's wrong here
max_tokens controls the maximum length of generated output, not how the model attends to or retains earlier instructions. Raising it will not restore adherence to policy details that appear in the system prompt. The contradiction likely stems from instruction dilution across a long context window, not from truncation of the model's reply, so this change does nothing to address the root cause.
- ✗
Move the policy text from the system prompt into the first user message of the conversation.
Why it's wrong here
Relocating the policy into the first user turn places it deep in the conversation history, where it is more likely to be diluted as turns accumulate, not less. The system prompt is the appropriate, persistent location for durable instructions. This change also removes the system-level authority that helps Claude treat the policy as binding rather than as conversational content.
- ✓
Restate the critical policy instructions in the system prompt and periodically re-inject a condensed reminder of them in recent turns.
Why this is correct
Long conversations dilute earlier instructions because attention is spread across many tokens. Keeping the policy in the system prompt preserves its authority, while periodically re-injecting a concise reminder near the current turn keeps the relevant constraints salient at generation time. This directly counteracts the drift observed after roughly 40 turns without discarding conversation history.
- ✗
Enable extended thinking so Claude reasons about the policy each time before responding.
Why it's wrong here
Extended thinking gives the model more internal reasoning budget for complex tasks, but it does not guarantee that distant policy text is weighted correctly in a long conversation. The drift is an attention and salience problem, not a reasoning-depth problem. Enabling it may increase latency and cost without resolving the contradiction of earlier policy statements.
About these practice questions
Courseiva writes every CCDV-F question from scratch — 257 in total, each with an explanation and a wrong-answer breakdown. None are copied from real exams or dumps. Learn why practice questions differ from exam dumps →
JA
Written and reviewed by Johnson Ajibi, MSc IT Security
Senior Network & Security Engineer · founder of Courseiva
Last reviewed September 2026 · checked against the official Anthropic exam blueprint
This CCDV-F practice question is part of Courseiva's free Anthropic certification practice question bank. Courseiva provides original exam-style practice questions with explanations, topic-based practice, mock exams, readiness tracking, and study analytics to help learners prepare for the CCDV-F exam.