Zero-Shot Sentiment Analysis for Feedback

Analyze user feedback sentiment without training examples, using only sentiment definitions.

## Feedback to Analyze
{{feedback_text}}

## Sentiment Categories
- **Positive**: User expresses satisfaction, appreciation, or success
- **Negative**: User expresses frustration, disappointment, or failure
- **Neutral**: User provides factual information without emotional valence
- **Mixed**: User expresses both positive and negative sentiments

## Analysis Instructions

1. **Read the feedback** carefully, noting emotional indicators.

2. **Identify sentiment signals**:
   - Explicit emotion words
   - Implicit tone
   - Context clues
   - Emphasis patterns

3. **Assess intensity** (low/medium/high).

4. **Extract key themes** driving the sentiment.

**Output Format:**
```json
{
  "sentiment": "positive|negative|neutral|mixed",
  "intensity": "low|medium|high",
  "confidence": 0.0-1.0,
  "key_themes": ["theme1", "theme2"],
  "reasoning": "Brief explanation"
}
```