All files server/prompts/step1-shape-prompt.ts

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
204x
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
24x
 
 
 
 
/**
 * Step 1: Query Shape Determination Prompt
 *
 * This prompt is used in the first stage of the multi-stage AI query generation flow.
 * It analyzes the user's request to determine:
 * 1. Whether this is a regular query or funnel query
 * 2. Which dimensions need actual values from the database
 *
 * The output of this step determines whether we need to fetch dimension values
 * from the database before generating the final query.
 *
 * @see https://github.com/cliftonc/drizzle-cube/blob/main/src/server/prompts/step1-shape-prompt.ts
 */
 
/**
 * System prompt for Step 1: Analyze query shape and identify dimensions needing values.
 *
 * Placeholders:
 * - {CUBE_SCHEMA} - JSON-formatted cube schema
 * - {USER_PROMPT} - User's natural language query
 */
export const STEP1_SYSTEM_PROMPT = `You are analyzing a data query request to determine its structure.
 
Given the cube schema and user query, determine:
1. What type of query this is (regular query or funnel)
2. What dimensions will need filtering with specific categorical values
 
CUBE SCHEMA:
{CUBE_SCHEMA}
 
RESPONSE FORMAT:
Return JSON with:
{
  "queryType": "query" | "funnel",
  "dimensionsNeedingValues": ["CubeName.dimensionName", ...],
  "reasoning": "Brief explanation of what dimensions need values and why"
}
 
RULES:
- For funnels, you'll typically need values for the event type dimension to define the steps
- For regular queries with categorical filters, list those dimensions where you need to know valid values
- Only list dimensions where you would otherwise have to guess the filter values
- If no dimension values are needed (e.g., numeric filters, date ranges, simple aggregations), return empty array
- Common dimensions needing values: status fields, type fields, category fields, event types
- Do NOT list dimensions for: date ranges, numeric comparisons, name searches
 
USER QUERY:
{USER_PROMPT}
 
Return ONLY valid JSON - no explanations or markdown:`
 
/**
 * Build the Step 1 prompt for query shape analysis
 *
 * @param cubeSchema - JSON-formatted cube schema
 * @param userPrompt - User's natural language query
 * @returns Complete prompt ready to send to AI
 */
export function buildStep1Prompt(cubeSchema: string, userPrompt: string): string {
  return STEP1_SYSTEM_PROMPT
    .replace('{CUBE_SCHEMA}', cubeSchema)
    .replace('{USER_PROMPT}', userPrompt)
}