Open-source backend for Kepler.gl, Dekart enables users to create and share live geospatial maps using SQL queries against data warehouses like BigQuery, Snowflake, and Postgres. It serves as a self-hosted alternative to commercial platforms such as CARTO and Felt, targeting engineers and data analysts who require greater control over their geospatial data and map creation workflows.

How It Works

Dekart functions as a self-hosted backend for Kepler.gl, built using Golang and React. It connects directly to various data warehouses, exposing an MCP (Map Creation Protocol) that allows agents (like Claude and Codex) to generate maps from data. This approach facilitates automated map creation and enhances accuracy in geospatial tasks, offering a flexible, code-driven alternative to proprietary solutions.

Quick Start & Requirements

To run Dekart, use the Docker command: docker run -p 8080:8080 dekartxyz/dekart. For optional agent capabilities, install the geosql skill (pip install geosql && geosql) and the Dekart CLI (pip install dekart-cli && dekart init). The default setup uses built-in SQLite and local file storage, requiring zero configuration. Production deployments can be configured with S3/GCS backups or a Postgres metadata backend. Deployment guides are available for Docker, Docker Compose, AWS/ECS, and Google App Engine.

Highlighted Details

• Connects to major data warehouses including BigQuery, Snowflake, and Postgres.
• Enables creation of live maps directly from SQL queries or via AI agents (Claude, Codex).
• Supports shareable map links and management of data access and sharing.
• Offers a single Docker container for straightforward deployment on-premises or in the cloud.

Maintenance & Community

The project is maintained by Volodymyr Bilonenko and has an active Slack community for support. Users are encouraged to star the repository for updates.

Licensing & Compatibility

Dekart is open-source under the GNU Affero General Public License Version 3 (AGPLv3) or any later version. Commercial licenses are also available. The AGPLv3 license imposes copyleft requirements, meaning derivative works must also be made available under the same license, which may have implications for closed-source commercial integration.

Limitations & Caveats

The agentic setup for map creation is presented as optional, suggesting core functionality may not rely on AI agents. The AGPLv3 license requires that any modifications or derivative works distributed must also be open-sourced under AGPLv3, potentially limiting its use in proprietary commercial applications without a commercial license.