query

name	query
description	Ask natural language questions about your data and get answers with evidence
argument-hint	[dataset-name] [your question]
risk	safe
disable-model-invocation	false
user-invocable	true
allowed-tools	Read, Write, Bash, Glob, Grep
model	claude-sonnet-4-6
context	fork
agent	general-purpose

10x Analyst — Data Query

Ask any question about your data in plain English and get a precise answer backed by numbers.

Overview

Loads data from input/<dataset>/, interprets your question, writes and executes the exact pandas query needed, and returns the answer with supporting evidence. No dashboard, no report — just the answer.

When to Use

• User asks a specific question about data: "What are our top 5 products?", "What's the average order value?", "Which customers churned?"
• User wants a quick answer without the full analysis pipeline
• User wants to explore data interactively with follow-up questions

Path Resolution

Parse $ARGUMENTS:

• First word: dataset name → reads from input/<dataset-name>/
• Remaining text: the user's question

Instructions

1. Parse $ARGUMENTS:
   • First word = dataset name (e.g., shopify-data)
   • Everything after = the question
   • Input path: input/<dataset>/
2. Find and load all data files at input/<dataset>/
3. Clean column names: df.columns = df.columns.str.strip().str.lower().str.replace(r'[^a-z0-9]+', '_', regex=True).str.strip('_')
4. If multiple files, join on common keys (columns ending in _id or named id)
5. Read the user's question and determine what pandas operations are needed
6. Write and execute a Python script that:
   • Loads the data from input/<dataset>/
   • Runs the query (groupby, filter, aggregate, sort, etc.)
   • Prints the result as a formatted table or value
7. Present the answer in this format:

## Answer

{Direct answer to the question with specific numbers}

### Supporting Data

{Table or list with the evidence}

### How This Was Computed

{1-2 sentences explaining the pandas operations used}

### Follow-Up Questions You Might Ask
- {Suggested question 1}
- {Suggested question 2}

Examples

# Specific metric
/10x-analyst:query shopify-data "What is the average order value?"

# Top-N question
/10x-analyst:query shopify-data "What are the top 10 products by revenue?"

# Segment question
/10x-analyst:query shopify-data "Which customer segment has the highest lifetime value?"

# Trend question
/10x-analyst:query shopify-data "Is revenue growing or declining month over month?"

Limitations

• One question at a time — for multi-question analysis use :analyze
• Cannot query external APIs or databases — file-based data only
• Complex multi-step questions may need to be broken down

---

Developed by 10x.in | 10x-Analyst v1.0.0