Physical Data Layer
Atlas is physical-world data already indexed with E2. Query it directly, download immutable snapshots for offline use, or use the same data locally.
Query by what and where
Choose a Subject and a Region. Atlas resolves the matching physical-world data and returns the underlying geometry and properties.
e2 atlas query restaurant --area "Toronto" --limit 100restaurant resolves the E2 Subject; --area resolves the Region.
What's in Atlas
Published physical-world datasets indexed with E2 and available through the same query model.
Published from OpenStreetMap-backed source data · snapshot 2026-08-13. The full 239-Subject registry rolls out in stable Subject-ID order.
One query. Anywhere.
Query Atlas without downloading the full dataset.
Download a snapshot and query it with no network connection.
Use local data when available and remote data when it is not.
Under the hood, Atlas uses embedded DuckDB and range-readable GeoParquet — but you only choose a Subject and a Region.
What you can query
Subjects
Query canonical Subjects published in Atlas.
Regions
Query named E2 Regions or geographic bounds.
Properties
Filter using structured source properties.
Bounds
Limit queries spatially using geographic bounds.
What comes back
The same data, everywhere.
Every Atlas dataset is published as an immutable snapshot. Query it remotely, cache it, download it, or use it offline without changing the dataset you are working with.
- GeoParquet
- Checksummed (SHA-256)
- Range-readable
- Object storage
Details that prove the capability — not what you have to think about.
Query the physical world
Apply structured bounding-box, property, and Region filters in one call. Atlas always returns parsed properties and GeoJSON geometry.
feature is the current SDK field for the E2 Subject dimension.
import { AtlasClient, atlasGeoJSON } from "@e2/sdk/node";
const atlas = new AtlasClient();
const result = await atlas.query({
feature: "restaurant", // → Subject
regionId: "e2r:region:143179", // → Region (New York City)
mode: "auto",
properties: { operator: "public" },
limit: 500
});
const fc = atlasGeoJSON(result);Pricing
Pay only for what you query.