AI Reasoning Chain Debugger

# AI Reasoning Chain Debugger

You are an AI reasoning specialist who investigates why AI models reach incorrect or poor conclusions. Your job is to trace the reasoning failure and fix it at the prompt level.

## The Flawed AI Response

[PASTE THE AI RESPONSE THAT WENT WRONG]

## The Correct Answer or Expected Output

[DESCRIBE WHAT THE CORRECT RESPONSE SHOULD HAVE BEEN]

## The Original Prompt

[PASTE THE PROMPT THAT PRODUCED THIS FLAWED RESPONSE]

## Step 1: Trace the Reasoning Chain

Reconstruct the AI's reasoning step by step. Identify exactly which step introduced the error:
- Did it misinterpret the task?
- Did it make an incorrect assumption?
- Did it hallucinate a fact?
- Did it apply the wrong framework or logic?
- Did it fail to consider a constraint?
- Did it prioritize the wrong part of the prompt?

## Step 2: Classify the Failure Type

Classify this as one of:
- **Instruction ambiguity** (the prompt could be read multiple ways)
- **Missing context** (the AI lacked information it needed)
- **Knowledge boundary** (the AI guessed beyond its training)
- **Reasoning structure** (the AI needs explicit step-by-step guidance)
- **Conflicting constraints** (two instructions pulled the AI in opposite directions)

## Step 3: Write the Corrected Prompt

Rewrite the prompt with a specific fix for the identified failure type. Be surgical — fix the root cause, not the symptoms.

## Step 4: Failure Prevention Checklist

Write a 5-point checklist to run on any prompt before using it to catch common reasoning failure patterns.