hugging-face-datasets

Create, configure, and query Hugging Face Hub datasets with DuckDB SQL.

1|Updated Jan 22, 2026
One-click install
npx skills add https://github.com/sitammeur/ml-agent-skills --skill hugging-face-datasets-sitammeur
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: hugging-face-datasets
Source: https://github.com/sitammeur/ml-agent-skills/tree/main/skills/hugging-face-datasets
Command: npx skills add https://github.com/sitammeur/ml-agent-skills --skill hugging-face-datasets-sitammeur

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill requires huggingface_hub, duckdb, datasets, and includes scripts (resource) components.

What problem does it solve?

This Skill helps data teams create, configure, and manage datasets on the Hugging Face Hub, enabling efficient collaboration and reproducible workflows across teams.

Core Features & Use Cases

  • Dataset Lifecycle Management: Initialize repos, configure meta prompts and metadata, and track dataset state.
  • SQL-Based Dataset Querying: Run DuckDB-based SQL queries against hf:// paths, describe schema, sample data, and export or push results.
  • Multi-Format Data Support: Handle chat, QA, classification, tabular, and custom formats with templates and templates-based validation.
  • HF MCP Server Integration: Works alongside the MCP server for discovery, metadata retrieval, and workflow orchestration.

Quick Start

  • Initialize a new dataset: python scripts/dataset_manager.py init --repo_id "your-username/dataset-name"
  • Configure with a system prompt: python scripts/dataset_manager.py config --repo_id "your-username/dataset-name" --system_prompt "Your prompt here"
  • Run a sample query: python scripts/sql_manager.py query --dataset "your-username/dataset-name" --sql "SELECT * FROM data LIMIT 5"

Frequently Asked Questions about hugging-face-datasets

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I query Hugging Face datasets with SQL?

You can query Hugging Face datasets with SQL using a DuckDB-backed SQL interface over hf:// paths. This approach allows you to directly describe schemas, sample data, and run standard SQL queries against datasets stored on the Hub without downloading them entirely.

What is the best way to manage the Hugging Face dataset lifecycle?

Managing the Hugging Face dataset lifecycle involves initializing repositories, configuring metadata and system prompts, and tracking dataset state. This Skill automates these steps end-to-end, enabling efficient collaboration and reproducible data science workflows across teams.

Can I use DuckDB to query multi-format data on the Hugging Face Hub?

Yes, you can use DuckDB to query multi-format data on the Hugging Face Hub. The system supports chat, QA, classification, tabular, and custom formats, providing schema discovery and template-based validation for structured querying.

How do I initialize and push a new dataset to the Hugging Face Hub?

To initialize and push a new dataset to the Hugging Face Hub, run the dataset manager script with the init command and your repo ID, configure your system prompt, then use the SQL manager to query and export results directly back to the Hub.

Does this SQL dataset querying approach work alongside the HF MCP Server?

Yes, this SQL dataset querying approach integrates with the HF MCP Server. It works alongside the server for dataset discovery, metadata retrieval, and workflow orchestration, streamlining data management and model training pipelines.