Annotate CSV

A Claude Code skill that adds AI-generated annotations to CSV files by processing each row with a prompt.

Features

• Classify, extract, or summarize data row-by-row
• Parallel processing for speed
• Use all columns or select specific context
• Inline prompts or prompt files for complex tasks

Installation

As a Claude Code Skill

git clone https://github.com/brook-miller/annotate-csv.git ~/.claude/skills/annotate-csv

Standalone

git clone https://github.com/brook-miller/annotate-csv.git
cd annotate-csv
cp .env.example .env
# Add your API key to .env

Setup

Create a .env file:

OPENAI_API_KEY=sk-...

Quick Start

# Classify sentiment
uv run annotate-csv.py "Classify sentiment: positive, negative, neutral" reviews.csv -c sentiment

# Extract information
uv run annotate-csv.py "Extract the product name mentioned" feedback.csv -c product

# Summarize
uv run annotate-csv.py "Summarize in one sentence" articles.csv -c summary

Options

Option	Description	Default
-o, --output	Output file path	<input>_annotated.csv
-c, --column	Annotation column name	annotation
--context	Columns to use (all or comma-separated)	all
--model	OpenAI model	gpt-5-mini
--parallelism	Parallel workers	10
--id-column	Column for progress display	-

Example

Input reviews.csv:

id,text
1,"Great product, love it!"
2,"Terrible experience, broke immediately"

Command:

uv run annotate-csv.py "Classify sentiment: positive, negative, neutral" reviews.csv -c sentiment

Output reviews_annotated.csv:

id,text,sentiment
1,"Great product, love it!",positive
2,"Terrible experience, broke immediately",negative

Requirements

• Python 3.11+
• OpenAI API key

License

MIT