Datamonger

Retrieve public research datasets with a recorded registry, verified artifact bytes, and consistent decoded values across languages.

Start with the Python guide, R guide, or Julia guide. Use the command-line client when you need verified raw artifacts. The site includes Python, R, and Julia API references and nine shared concepts. Each language client and the CLI also has an authored guide.

Reproducible data access

A registry record fixes the source bytes, decoding recipe, expected shape, and canonical digest of decoded values. Record the dataset version and the strong registry selector with your analysis. A bare registry release name is a lookup, not a cryptographic pin.

All clients bundle the stable 1.1.0 registry snapshot. A bundled index works offline, but an uncached artifact requires access to its upstream source. Dataset artifacts are not bundled.

Understand the guarantees

Artifact hashes check bytes against the selected registry. Canonical digests check decoded logical values. Neither authenticates a selector obtained from an untrusted source nor grants permission to use a dataset.

Read the trust model and the revision 1 specification for the contracts and their limits. The contributor guide explains how to work on clients and registry records.