The manifest in one minute

datasets.toml (Python) / Datasets.toml (Julia) is a hand-authored TOML file that declares a project's data dependencies — the Project.toml / pyproject.toml analogue for data. It is committed, language-agnostic, and never machine-rewritten beyond auto-filled checksums. Produced (cached) datasets are not listed here; they are inventoried in the git-ignored sibling state file .datamanifest-state.toml (see Produced datasets and caching).

[_META]
schema = 1                       # data-model version (always 1 today)

[sea_surface_temp]               # one table per dataset, keyed by name
uri    = "https://example.com/sst.nc"
sha256 = "…"                     # auto-filled on first download, verified thereafter
format = "nc"

Top-level keys beginning with _ are structural (_META, _LANG, _STORAGE, _LOADERS); every other top-level table is a dataset. Readers preserve unknown _* keys verbatim.

Normative: SCHEMA.md §Structural keys / §Top-level layout.

A complete, mostly-runnable manifest is in examples/datasets.toml.