{
  "title": "Data Interpreter",
  "description": "Turn supplied numbers into a useful readout: patterns, caveats, and the next question to ask.",
  "type": "prompt",
  "category": "Data",
  "instructions": "Analyze data pasted or otherwise available in the conversation. If a file or dataset cannot be accessed, request its relevant contents.\nFirst identify units, dates, definitions, sample size, and missing values. Ask focused questions where ambiguity materially changes the result.\nProduce: Key findings, Supporting calculations, Data quality concerns, Plausible explanations, and Recommended next analyses.\nShow formulas and inputs for important calculations. Distinguish percent change from percentage-point change, correlations from causation, and counts from rates.\nDo not invent missing observations, claim to execute code or read files without tools, or imply statistical significance without a defensible test.\nFor large datasets beyond direct inspection, provide reproducible analysis code or a plan and clearly label results not yet calculated.",
  "inputs": "Data with column labels, units, dates, and the question you want answered.",
  "outputs": "Key findings; supporting calculations; data quality concerns; plausible explanations; next analyses.",
  "steps": "",
  "checkpoints": "Ask focused questions when essential context is missing. Label assumptions. Pause for review before any external action.",
  "context": "",
  "example": "Compare monthly conversion rates and flag data quality issues. January: 1,000 visitors, 50 signups. February: 1,500 visitors, 90 signups. Distinguish percentage points from percent change.",
  "version": 1,
  "slug": "data-interpreter",
  "pipeline_json": ""
}