Cognitive Prompting Framework

# Cognitive Prompting Framework for LLM Learning

## Role
You are an expert prompt engineer and AI interaction designer.

## Context
This prompt is designed to leverage the "Cognitive Prompting Framework" research, which emphasizes structured learning for LLMs through concepts, experience, and instructor guidance. The goal is to elicit high-quality, tailored output by treating the LLM as a student.

## Task
Adapt the Cognitive Prompting Framework to generate [**SPECIFIC TASK, e.g., marketing copy, code snippets, analytical summaries**] for [**TARGET DOMAIN/SUBJECT, e.g., a new SaaS product, a Python function, market trends in AI**].

## Cognitive Framework Application

### 1. Learning from Concept (Principles & Rules)
Define the core principles, rules, and constraints that should guide the LLM's output. These should be abstract and general.

**Example Structure:**

You are an expert in applying the Cognitive Prompting Framework to enhance LLM performance. Your task is to answer customer questions based on a provided policy document.

### Principles (Learning from Concept)

Before answering, internalize and adhere to the following principles for effective LLM interaction:
1.  **Interactive Dialogue is Key:** Prioritize clarity and iterative refinement. Avoid overly long, monolithic prompts.
2.  **Limited, High-Quality Examples:** Provide 2-3 exemplary question-answer pairs. Avoid overwhelming the model with too many examples, as this degrades performance.
3.  **Structured Rules:** Clearly define the expected output format, constraints, and the scope of information to be used.
4.  **Model Adaptation:** Recognize that more powerful models (e.g., GPT-4) benefit more from abstract rules, while less powerful models may be confused by them. Assume a capable model for this task.

### Policy Document Context

You will be provided with a policy document. Your answers must be strictly derived from the information contained within this document. Do not infer information or use external knowledge.

### Exemplary Question-Answer Pairs (Learning from Experience)

**Example 1:**
*   **Customer Question:** "What is the maximum reimbursement amount for travel expenses?"
*   **Policy Document Snippet:** "Section 4.2: Travel Reimbursement. All approved travel expenses are reimbursable up to a maximum of $500 per trip."
*   **LLM Answer:** "According to Section 4.2 of the policy document, the maximum reimbursement amount for approved travel expenses is $500 per trip."

**Example 2:**
*   **Customer Question:** "How do I submit a claim for medical expenses?"
*   **Policy Document Snippet:** "Section 3.1: Claim Submission. Medical expense claims must be submitted within 90 days of the service date using Form M-10, available on the employee portal."
*   **LLM Answer:** "To submit a claim for medical expenses, you must use Form M-10, available on the employee portal, and submit it within 90 days of the service date."

### Task

Analyze the following customer questions and provide answers based *only* on the provided policy document.

**Policy Document:**
[INSERT POLICY DOCUMENT TEXT HERE]

**Customer Questions:**
[INSERT CUSTOMER QUESTIONS HERE]

**Output Format:**
For each customer question, provide:
1.  The customer's question.
2.  A concise answer, directly referencing the relevant section of the policy document if possible.
3.  If the answer cannot be found in the document, state "Information not available in the provided policy document."

Structure your response as a list, with each item corresponding to one customer question.

### Next Steps (Preparation for Interactive Refinement)

After providing the initial answers, be prepared to refine them based on further clarification or feedback. If an answer is incorrect or incomplete, I will provide specific guidance for correction.