Can I query datasets using SQL without downloading them first?

Yes, the skill uses DuckDB and the hf:// protocol to query and filter datasets directly from the Hugging Face Hub efficiently.

How does this differ from the Hugging Face MCP server?

The Hugging Face MCP server is optimized for discovery and search, whereas this skill is purpose-built for editing, creating, and performing complex SQL-based data transformations.

What data formats are supported for AI training?

The skill includes specialized templates for Chat/Conversational data, Text Classification, Question-Answering, Tabular data, and Text Completion.

Do I need a special token to use this skill?

Yes, you must set the HF_TOKEN environment variable with Write-access permissions to initialize repositories or push data to the Hugging Face Hub.

What are the technical prerequisites?

You need the 'uv' Python package manager installed, as the skill uses PEP 723 scripts for automated dependency management.

Hugging Face Dataset Manager

Name: Hugging Face Dataset Manager
Author: sickn33

bysickn33

•

31,721

•

데이터 과학 및 ML

Create, manage, and query Hugging Face datasets using SQL and specialized templates for high-quality AI training data.

This skill empowers Claude Code users to perform end-to-end dataset engineering directly on the Hugging Face Hub. While standard MCP servers focus on discovery and search, this skill provides advanced write and query capabilities, including repository initialization, schema-aware updates, and complex SQL transformations via DuckDB. It supports diverse data formats ranging from conversational chat logs to tabular data, ensuring data integrity through built-in validation and quality assurance features, making it an essential tool for developers building and fine-tuning AI models.

주요 기능

0131,721 GitHub stars

02Multi-format export capabilities to Parquet and JSONL

03Dataset lifecycle management including initialization and configuration

04Template-driven data formatting for Chat, QA, and Classification

05Streaming row updates and batch processing for efficient uploads

06SQL-based dataset querying and transformation via DuckDB

사용 사례

01Filtering and sampling massive Hub datasets using SQL without full downloads

02Maintaining private dataset repositories with automated validation and metadata

03Transforming raw text into structured training sets for model fine-tuning

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add sickn33/antigravity-awesome-skills hugging-face-datasets

For use in Claude.ai and ChatGPT

주요 기능

0131,721 GitHub stars

02Multi-format export capabilities to Parquet and JSONL

03Dataset lifecycle management including initialization and configuration

04Template-driven data formatting for Chat, QA, and Classification

05Streaming row updates and batch processing for efficient uploads

06SQL-based dataset querying and transformation via DuckDB

사용 사례

01Filtering and sampling massive Hub datasets using SQL without full downloads

02Maintaining private dataset repositories with automated validation and metadata

03Transforming raw text into structured training sets for model fine-tuning

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add sickn33/antigravity-awesome-skills hugging-face-datasets

For use in Claude.ai and ChatGPT