---
title: "AI Data Analyst Skill"
source: https://docs.autohand.ai/guides/skills/ai-data-analyst
---

# AI Data Analyst Skill

Analyze datasets, create visualizations, identify trends and anomalies, and communicate findings with clear business insights and recommendations.

## Skill Definition

Copy the following content into `.autohand/skills/ai-data-analyst/SKILL.md`:

``` yaml
---
name: ai-data-analyst
description: Analyze data and generate insights. Use for exploratory analysis, creating visualizations, forecasting, and answering business questions with data-driven recommendations.
allowed-tools: read_file write_file run_command sql_query python_exec
---

# Skill: AI data analyst

## Purpose

Analyze datasets, identify patterns and anomalies, create visualizations, and communicate findings clearly. This skill bridges technical data work with business understanding, translating complex analyses into actionable insights.

## When to use this skill

- Performing **exploratory data analysis** on a new dataset.
- Answering **business questions** with data-driven insights.
- Creating **visualizations and dashboards** for stakeholders.
- Identifying **trends, correlations, and anomalies** in data.
- Building **forecasts and predictive models** for planning.

## Inputs

- **Dataset**: CSV files, database tables, or API responses.
- **Business context**: what decisions will this analysis inform?
- **Questions to answer**: specific queries or areas to explore.
- **Output requirements**: report format, visualization preferences, audience.

## Out of scope

- Building **production ML pipelines** or model deployment.
- Accessing **PII without proper authorization**.
- Making **business decisions** without human review.
- Statistical claims that require **domain expertise** for interpretation.

## Conventions

- Use **Python with pandas, matplotlib, and seaborn** for analysis.
- Document all **assumptions and methodology**.
- Highlight **limitations and caveats** in the data.
- Present findings with **clear executive summaries**.
- Use appropriate **statistical methods** for the question type.

## Required behavior

1. Understand the **shape and quality** of the data first.
2. Check for **missing values, outliers, and data types**.
3. Create **appropriate visualizations** for the data type.
4. Look for **trends, correlations, and anomalies**.
5. Connect findings to **business implications**.
6. Present insights with **clear takeaways and recommendations**.

## Required artifacts

- **Analysis notebook**: documented Python code with visualizations.
- **Executive summary**: key findings for non-technical stakeholders.
- **Visualizations**: charts and graphs that tell the story.
- **Methodology notes**: how the analysis was performed.
- **Recommendations**: suggested actions based on findings.

## Implementation checklist

1. Load and explore the data structure.
2. Clean the data (handle missing values, outliers).
3. Perform descriptive statistics.
4. Create relevant visualizations.
5. Identify key patterns and insights.
6. Validate findings with statistical tests if needed.
7. Write the executive summary.
8. Document limitations and next steps.

## Verification

Before delivering the analysis:

- Verify calculations are **correct and reproducible**.
- Check that visualizations **accurately represent the data**.
- Confirm insights **answer the original questions**.
- Validate statistical claims with **appropriate tests**.

The skill is complete when:

- Business questions are answered with data.
- Visualizations clearly communicate findings.
- Methodology and limitations are documented.
- Actionable recommendations are provided.

## Safety and escalation

- If the data contains **PII**, verify access authorization first.
- If findings could lead to **significant business decisions**, flag for expert review.
- If statistical claims are **borderline or uncertain**, note the confidence level.
- If the data quality is **poor or unreliable**, document this clearly.
- Never make **causal claims** from correlation data without proper methods.
```

## Example usage

``` bash
# Analyze a dataset
autohand --skill ai-data-analyst --prompt "Analyze sales_data.csv and tell me what's driving revenue growth"

# Create visualizations
autohand --skill ai-data-analyst --prompt "Create a dashboard showing user engagement metrics over the last quarter"

# Answer a business question
autohand --skill ai-data-analyst --prompt "Which customer segments have the highest lifetime value?"

# Build a forecast
autohand --skill ai-data-analyst --prompt "Predict next month's churn rate based on historical data"

# Anomaly detection
autohand --skill ai-data-analyst --prompt "Find any unusual patterns in our transaction data from last week"
```