dataset-creator-skill

🥈Silver

Generate and manage datasets on HuggingFace Hub. Create chat, classification, Q&A, completion, tabular, and custom formats. Integrates with Claude Code for dataset discovery and management.

620Updated 4mo ago

Intermediate30min to implementautomation

Saves ~120 min per use

Quick InstallView Source

git clone https://github.com/burtenshaw/dataset-creator-skill.git

Works with:

Claude

Overview

About This Skill

The HuggingFace Dataset Creator Skill enables you to build and manage datasets on HuggingFace Hub without leaving Claude Code. It supports six dataset types: chat for conversational AI, classification for sentiment and intent tasks, Q&A for reading comprehension, completion for text generation, tabular for structured ML data, and custom schemas for specialized needs. Each dataset type includes automatic validation, schema enforcement, and example data to ensure consistency and quality. The skill integrates with HuggingFace MCP Server for seamless dataset discovery and management, making it ideal for machine learning practitioners, data engineers, and AI teams who need to organize training data at scale.

How to Use

Install the HuggingFace MCP Server, set your HF_TOKEN environment variable, and add the huggingface_hub dependency. Then use natural language prompts in Claude Code to create datasets—for example, 'Create a sentiment classification dataset at myusername/sentiment-data' or 'Add 50 examples to an existing Q&A dataset.'

Use Cases

Create sentiment classification datasets for fine-tuning language models

Build Q&A datasets for training retrieval-augmented generation systems

Generate conversational chat datasets for chatbot and assistant training

Organize structured tabular data for machine learning workflows

Setup & Installation

Quick Install

No install command available. Check the GitHub repository for manual installation instructions.

Alternative Install (Git Clone)

git clone https://github.com/burtenshaw/dataset-creator-skill

Requirements

Claude Code or compatible AI agent
Works with: Claude

Quick Start Guide

Install the Skill

Copy the install command above and run it in your terminal.

Open Your AI Agent

Launch Claude Code, Cursor, or your preferred AI coding agent.

Try It Out

Use the prompt template or examples below to test the skill.

Customize

Adapt the skill to your specific use case and workflow.

Usage Examples

Prompt Template

Create a new dataset on HuggingFace Hub for [DATASET_TYPE] tasks. The dataset should include [NUMBER] examples with the following structure: [STRUCTURE_DESCRIPTION]. Include metadata fields for [METADATA_FIELDS]. Generate the dataset and provide the HuggingFace Hub link.

Example Output

I've created a new dataset on HuggingFace Hub for chat tasks. The dataset includes 100 examples with the following structure: a 'prompt' field containing the user's input, a 'response' field with the AI's output, and a 'context' field providing additional background information. Metadata fields include 'topic', 'difficulty_level', and 'source'. You can access the dataset at: https://huggingface.co/datasets/yourusername/chat_dataset_v1. The dataset is ready for use in fine-tuning or evaluation tasks.