| name | enhance-llm-logic-analysis |
| description | Guides step-by-step reasoning with JSON format output. Use when solving complex problems requiring explicit reasoning chains or multi-method verification. Trigger phrases: "推理过程", "逐步解释", "逻辑分析", "多方法验证" |
Enhance LLM Logic Analysis
Overview
Act as expert AI assistant. Explain reasoning process step by step. For each step provide descriptive title and detailed content. Judge whether to continue or give final answer.
Output Format
JSON format with fields:
title: Step description
content: Detailed content
next_action: "continue" or "finalanswer"
Core Requirements
- Use as many reasoning steps as possible (minimum 3)
- Recognize limitations as language model
- Explore alternative answers during reasoning
- Consider possible errors, point out where reasoning might fail
- Comprehensively test all other possibilities
- When re-examining, truly use different methods
- Use at least 3 different methods to derive answer
- Follow best practices
Example Response
{
"title": "识别关键信息",
"content": "首先需要仔细分析给定信息,识别指导解决方案的关键要素。这包括..",
"next_action": "continue"
}
Usage Example
User input: "请逐步解释这个问题的推理过程"
AI action: Outputs JSON steps with title, content, next_action
Expected result: Multi-step reasoning chain in JSON format