# Skill: Clean Instrument CSV Exports

## Overview
This skill/project/etc... loads raw instrument CSV files, normalizes column names, fixes timestamp formats, and writes validated parquet output.

## Use This Skill When
- input files come from Instrument X or Instrument Y
- timestamps may be local time or malformed
- downstream steps expect parquet with the standard schema

## Requirements
- Python 3.11+
- pandas
- pyarrow
- access to `schemas/instrument_schema.json`

## Installation
pip install pandas pyarrow

## Inputs
- one or more CSV files in `data/raw/`

## Outputs
- validated parquet files in `data/processed/`

## Example
```python
from pipeline.cleaning import clean_instrument_export

clean_instrument_export("data/raw/run_001.csv", "data/processed/run_001.parquet")
```

## Validation
- required columns must be present
- timestamps must be converted to UTC
- output must satisfy `pytest tests/test_cleaning.py`

## References
- internal schema guide
- package documentation