What is E2?
An open-source, local-first stack for places, search, routing, and spatial compute over open data.
Documentation navigation
Start with E2
SDK guides
CLI guides
MCP guides
Applications
Framework quickstarts
The problem
Location is one of the most locked-down corners of software. Proprietary map APIs meter every request, price by call, and reserve the right to change terms. Their terms of service routinely restrict caching, storage, and redistribution, so every place name, road shape, and routing result you display is rented rather than owned. Migrating means rewriting your app against a different API with different data.
The underlying data is mostly open already. OpenStreetMap and related catalogs describe hundreds of millions of features. E2 exists to make that data as easy to query as a hosted API, without the lock-in.
What E2 is
E2 is an open-source stack for places, search, routing, and spatial compute built on open data. The same operations — search, routes, matrices, isochrones, spatial analysis — run against a local SQLite index on your machine or against hosted APIs. You switch between the two with a mode flag, not a rewrite.
The pieces
| Piece | What it does |
|---|---|
@embed-earth/sdk-test | JavaScript and TypeScript SDK for search, routing, compute, storage, catalogs, and cells. |
@embed-earth/cli-test | The e2 CLI for search, routing, compute, cache management, and local indexing from the terminal. |
@embed-earth/mcp-test | An MCP server that exposes E2 tools to AI agents over standard input and output. |
| Hosted platform | APIs for search, routing, tiles, and compute when you want a managed endpoint. |
| Open data catalog | Published feature and region datasets you can download and index locally. |
| Map tiles | Raster and vector tiles that render with open libraries. |
Local-first architecture
The SDK, CLI, and MCP server all read and write the same local SQLite data. Snapshots of published datasets live in a cache directory you control. Queries run in remote mode (fetch published data), offline mode (cached data only), or auto mode (cache first, network when needed). Offline mode works with no network at all.
Your data stays yours. Queries against your local index never leave your machine, and the data files are ordinary SQLite you can copy, back up, inspect, or drop.
Who it is for
| You are | E2 gives you |
|---|---|
| A developer replacing Google or Mapbox APIs | Drop-in search, routing, and compute with predictable local or hosted pricing and no storage restrictions. |
| A data team | Open catalogs, local SQLite indexes, and spatial compute primitives that run next to your pipelines. |
| Building AI agents | An MCP server with typed tools for search, routing, compute, and catalogs. |