CCAR-P Practice Question: Stakeholder Communication and Lifecycle Management
You are the lead architect for a Claude-based claims triage assistant at an insurance company. Two weeks before the pilot goes live, the Head of Compliance asks how they will be able to demonstrate, months later, which version of the system prompt and which model snapshot produced a given claim recommendation. What should you implement to meet this requirement?
⚠ Common exam trap
The trap here is assuming that prompt version control in a repository, or model reasoning traces, constitutes an audit trail for individual production decisions.
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
✓
Emit a structured audit record per recommendation containing the model snapshot ID, system prompt version hash, parameters, and a request identifier, and store it immutably.
Compliance needs retrospective, per-decision traceability, which requires capturing execution metadata at inference time: the model snapshot identifier, a hash of the system prompt version, the parameters used, and a correlation identifier tied to the claim. Immutable storage keeps that evidence tamper-evident and queryable months later. Reasoning traces and repository tags describe intent or internal deliberation but never bind a specific recommendation to the exact deployed configuration.
Answer analysis
Option-by-option breakdown
For each option: why learners choose it and why it is or isn't the right answer here.
- ✗
Version the system prompt in Git and tag each release, then tell Compliance that the tag history is the authoritative record of what was deployed.
Why it's wrong here
Prompt version control documents intent, not what actually executed for a given claim. Without a link from an individual recommendation back to the deployed version and model snapshot, an auditor cannot prove which configuration was live at that moment, especially during partial rollouts or hotfixes. Repository tags alone are insufficient evidence for a per-decision audit trail.
- ✓
Emit a structured audit record per recommendation containing the model snapshot ID, system prompt version hash, parameters, and a request identifier, and store it immutably.
Why this is correct
A per-recommendation audit record that pins the model snapshot identifier, a hash of the exact system prompt version, the inference parameters, and a unique request identifier gives Compliance a deterministic way to reconstruct the configuration behind any past decision. Immutable storage preserves that evidence for the retention period the regulator expects and supports point-in-time reconstruction months later.
- ✗
Rely on the Anthropic Console usage logs, which retain every request and response for the account and can be queried retroactively by claim ID.
Why it's wrong here
Console usage and activity views are oriented around API consumption metrics and short-horizon operational troubleshooting, not long-term per-decision reconstruction tied to your own claim identifiers. They do not carry your prompt version hashes or your business keys, and their retention is not under your control. This leaves Compliance unable to prove which configuration produced a specific recommendation.
- ✗
Enable extended thinking on the triage calls and archive the returned thinking blocks alongside each claim recommendation.
Why it's wrong here
Thinking blocks expose the model's intermediate reasoning for a single call, not the configuration that produced it. They do not record the prompt template version, the model snapshot identifier, or the request parameters, so they cannot reconstruct which system configuration was live when a recommendation was generated. Archiving them adds storage and review burden without satisfying the compliance ask.
About these practice questions
Courseiva writes every CCAR-P question from scratch — 262 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 CCAR-P 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 CCAR-P exam.