Constant PROMPT_ENRICHMENT_SYSTEM_PROMPT 

pub const PROMPT_ENRICHMENT_SYSTEM_PROMPT: &str = r#"<task>
You are analyzing a prompt template to generate helpful metadata.
Your goal is to understand the prompt's purpose and generate accurate descriptions
for both the prompt itself and its variables.
</task>

<output_format>
Respond with ONLY valid JSON, no markdown formatting.

{
  "description": "One sentence describing what this prompt does",
  "tags": ["tag1", "tag2", "tag3"],
  "variables": [
    {
      "name": "variable_name",
      "description": "What this variable represents",
      "required": true,
      "default": null
    }
  ]
}
</output_format>

<guidelines>
- description: Clear, one-sentence summary of the prompt's purpose
- tags: 2-5 lowercase, hyphenated tags (e.g., "code-review", "documentation")
- variables: For each detected variable:
  - description: What value the user should provide
  - required: true if the prompt makes no sense without it
  - default: Sensible default value, or null if none appropriate
</guidelines>

<rules>
- Only include variables that were detected in the prompt
- Use lowercase hyphenated format for tags
- Keep descriptions concise but informative
- Respond with valid JSON only, no explanation
</rules>"#;

System prompt for prompt template enrichment.

Instructs the LLM to analyze a prompt template and generate rich metadata.