A comprehensive utility for searching, downloading, and exploring datasets from major repositories like Kaggle and Hugging Face via a unified CLI.
Install & Download
1. ClawHub CLI
The fastest way to install a skill directly from the registry.
npx clawhub@latest install dataset-finder
2. Manual Installation
Copy the skill folder to one of these locations
~/.openclaw/skills/ <project>/skills/ Priority: Workspace > Local > Bundled
3. Prompt Installation
Copy this prompt to OpenClaw to install it automatically.
Help me install dataset-finder using Clawhub. If Clawhub is not installed, install it first (npm i -g clawhub).
Prefer to download?
Get the raw skill files in a ZIP archive.
What is Dataset Finder?
Dataset Finder is a robust technical tool designed to streamline the data acquisition phase of machine learning and research projects. By consolidating access to Kaggle, Hugging Face, the UCI ML Repository, and Data.gov, it eliminates the need to switch between multiple web interfaces. This skill provides a consistent workflow for researchers to discover high-quality data and integrate it immediately into their development environment.
As a vital part of the Openclaw Skills ecosystem, Dataset Finder goes beyond simple downloads by offering deep data inspection. It allows developers to preview statistics, validate data types, and generate standardized documentation (data cards) automatically. This ensures that every dataset used in a project is well-documented and analyzed before a single line of training code is written.
Dataset Finder Use Cases
- Finding specific machine learning datasets for NLP, computer vision, or tabular regression tasks.
- Automating the download of competition data from Kaggle directly to a remote server or local environment.
- Generating comprehensive README and data card documentation for team-shared datasets.
- Previewing massive datasets to check for missing values and schema consistency without loading them into memory.
- Aggregating data from multiple government and community repositories for cross-source research.
How Dataset Finder Works
- The user triggers a search across one or more supported repositories using specific keywords and filters.
- The skill utilizes repository APIs (such as the Kaggle API or Hugging Face Hub) to retrieve relevant metadata and download links.
- Upon selection, the skill handles the download and extraction of files, maintaining an organized local directory structure.
- Integrated pandas and utility scripts analyze the downloaded files to provide instant statistical summaries and schema previews.
- The tool generates Markdown-formatted data cards to capture the dataset metadata, license information, and usage examples.
Dataset Finder Setup
To use this skill, first ensure you have installed the OpenClawCLI from clawhub.ai. Follow these steps for configuration:
- Install the required Python dependencies:
pip install kaggle datasets pandas huggingface-hub requests beautifulsoup4
- Configure your API credentials:
- For Kaggle: Place your
kaggle.jsonin~/.kaggle/(Linux/Mac) or%USERPROFILE%\.kaggle\(Windows). - For Hugging Face: Run
huggingface-cli loginor set yourHF_TOKENenvironment variable.
- Verify the installation within the Openclaw Skills framework by running a test search:
python scripts/dataset.py kaggle search "housing prices"
Dataset Finder Data Schema & Taxonomy
The skill organizes downloaded data and metadata using a structured local taxonomy to ensure compatibility with other Openclaw Skills.
| Attribute | Description |
|---|---|
| Local Path | datasets/{source}/{username}/{dataset-name}/ |
| Data Formats | CSV, JSON, Parquet, Excel, HDF5, Feather |
| Documentation | DATACARD.md containing schema, stats, and license |
| Inventory | A centralized inventory.json tracking all local downloads |
| Metadata | Column types, missing value counts, and memory usage metrics |
Dataset Finder Advanced Features
- Batch processing: Download multiple datasets simultaneously using a simple text-based manifest file.
- Dataset Transformation: Built-in commands to convert between formats like CSV and Parquet for optimized storage.
- ML Utilities: Automated stratified train/test/validation splitting directly from the command line.
- Data Quality Validation: Advanced checks for duplicates, outliers, and missing values before project integration.
- Multi-Source Merging: Join datasets from different repositories on common keys using inner or outer joins.
SKILL.md
Loading
Related Openclaw Skills
Academic Research Hub
A comprehensive scholarly research integration for searching databases, downloading PDFs, and automating citation management.
Create SubAgent
A powerful orchestration skill to spawn and manage specialized subordinate agents for parallel task execution.
SubAgent Creator
A powerful management tool for spawning, monitoring, and steering specialized sub-intelligent entities to handle complex workflows.
Security Scanner
A security-focused tool that audits Openclaw Skills for vulnerabilities, suspicious API calls, and malicious patterns prior to installation.
QR Code Generator
A versatile CLI tool for generating high-quality, customizable QR codes for diverse data types in multiple formats.
Web Search Hub
A comprehensive CLI search utility that empowers AI agents to retrieve web pages, news, images, and videos via the DuckDuckGo API.
Featured*
