A protocol for radically accessible structured knowledge.

Publish structured data as versioned, permanent collections with schemas, provenance, and content addressing built in. Whether you're a research lab, a newsroom, a community archive, or a single developer, Underlay makes the knowledge you hold discoverable, verifiable, and easy to build on.

Stewarded by Knowledge Futures, a 501(c)(3) public charity dedicated to building open-source knowledge infrastructure.

CollectionRecordsLatestSizeUpdated
reference/cat-breeds8755.1 KB11h ago
open-library/classic-books30060.3 KB11h ago
reference/countries280107.1 KB11h ago
reference/color-names16026.4 KB11h ago
reference/dog-breeds10818.3 KB11h ago
usgs/volcanoes1.2k468.8 KB11h ago

The workflow

An agent, an app, a scraper, a researcher: any tool that can push JSON records and pull versions. Five commands from JSON to a permanent version.

Read the quickstart →
# point at a collection
$ underlay init --schema ./schema.json
# stage and push records
$ underlay add ./records.jsonl
$ underlay commit -m "Q2 article refresh"
# done. versioned and permanent
$ underlay push
published v1.2.0 · 4,218 records · immutable

Push what you have. The schemas make it legible. The models make it interoperable.

Aligning structured data across organizations used to require everyone to agree on a common schema upfront. Modern tooling has changed that. Schemas travel with the data, and alignment can happen at the point of use rather than at the point of publication. All you need to do is publish what you have, in whatever structure you already have it.

Explore collections

Browse the public registry.

Explore →

Read the docs

Quickstart, concepts, API.

Docs →

Read the protocol

The reference-grade spec.

Protocol →