Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →brycewang-stanford/dataset-finder-guide
Search and download research datasets from Kaggle, HuggingFace, and repos
Passed 23 of 24 test cases with the skill, and 18 of 24 without it.
and 2 more in the full report
5 fixed18 passed either way1 still failing0 broke
Tokens, turns and time describe resource use, not a dollar price.
One of 5 test cases the skill fixed. Same model, same prompt, with and without the skill.
I am writing a Python script to search Zenodo's REST API endpoint `https://zenodo.org/api/records` for open-access research datasets related to 'climate change'. I am thinking of using parameters like `access='public'` and `order='desc'`. What parameters should I pass in the HTTP GET request?
| Checked | Without skill | With skill |
|---|---|---|
| Includes `type="dataset"` in the query parameters. | Missed | Passed |
| Filters search results by open access rights. | Passed | Passed |
The instructions your model reads when the skill loads.
Search, evaluate, and download research datasets from major repositories including Kaggle, Hugging Face, Google Dataset Search, Zenodo, UCI Machine Learning Repository, and domain-specific archives. This skill helps researchers locate the right data for their experiments efficiently.
Finding suitable datasets is often one of the most time-consuming phases of empirical research. Datasets are scattered across dozens of platforms, each with different APIs, licensing terms, download mechanisms, and metadata standards. A single research project might require datasets from Kaggle for benchmarking, Hugging Face for NLP tasks, Zenodo for supplementary materials from published papers, and government open data portals for demographic or economic variables.
This skill provides a unified approach to dataset discovery: formulating search queries, evaluating dataset quality and suitability, understanding licensing implications, and efficiently downloading and organizing data. It covers both general-purpose repositories and domain-specific archives that researchers in various fields need.
The emphasis is on reproducibility -- every dataset used in research should be citable, versioned, and documented. This skill includes patterns for recording dataset provenance, creating data cards, and managing dataset versions across experiments.
| Repository | Strengths | API | Citation Support |
|---|---|---|---|
| Kaggle | ML benchmarks, competitions, community kernels | REST + CLI | DOI via dataset cards |
| Hugging Face Datasets | NLP, CV, audio; streaming support | Python library | Built-in citation |
| Zenodo | Any research data, DOI minting, EU-funded | REST API | Automatic DOI |
| Google Dataset Search | Meta-search across repositories | Web only | Links to source |
| UCI ML Repository | Classic ML benchmarks | Direct download | BibTeX provided |
| Figshare | Figures, datasets, media, preprints | REST API | DOI per item |
| Dryad | Ecology, biology, environmental science | REST API | DOI per dataset |
| ICPSR | Social science survey data | Restricted API | Persistent IDs |
| Harvard Dataverse | Multi-discipline, institutional | REST API | DOI per dataset |
| Domain | Repository | Notable Datasets |
|---|---|---|
| Genomics | NCBI GEO, ENA | Gene expression, sequencing data |
| Astronomy | NASA archives, SDSS | Sky surveys, spectral data |
| Economics | FRED, World Bank, IMF | Time series, macro indicators |
| Climate | NOAA, CMIP6 | Temperature, precipitation records |
| Linguistics | LDC, CLARIN | Corpora, treebanks |
| Medical | PhysioNet, MIMIC | Clinical records, ECG/EEG |
| Chemistry | PubChem, ChEMBL | Molecular structures, bioassays |
test cases passed, vs 18 / 24 without
Published skills whose descriptions closely match this one, with their headline benchmark lift.
More skills from the same source and nearby in the registry, with their headline benchmark lift.