drt
github.com
- Free plan
- Windows
- macOS
- Linux
- Self-hosted
What drt is
drt is a free reverse ETL tool for moving modeled warehouse data into team tools, with sync definitions kept in YAML and versioned in Git. Users install `drt-core` with pip and run syncs from a terminal using `drt run`. A dry run with a diff previews rows slated for addition, update, or deletion before writing. The project lists 13 sources and 34 destinations, including BigQuery, Snowflake, Postgres, Slack, and Salesforce. It can read dbt projects and use `ref()` models to follow model lineage. Listed orchestration integrations include Dagster, Prefect, Airflow, and a dbt manifest reader; a GitHub Action supports scheduled or event-triggered syncs. An optional MCP server lets compatible clients list and run syncs, validate configuration, and inspect status. The project says data passes directly from the user’s warehouse to the destination rather than through a drt-hosted intermediary. Telemetry is off by default; when enabled, the project says it collects sync metadata but not sync names, SQL, destination URLs, credentials, or project paths. The core is self-hosted and Apache 2.0 licensed. RBAC, audit logging, plugins, and managed cloud hosting are outside drt-core.
Who it suits
It suits teams that manage data work with dbt and Git and want reverse ETL pipelines maintained in their repository. It is a self-hosted, terminal-driven option rather than managed cloud hosting.
Good to know, and what to watch
- Syncs are defined in YAML and versioned in Git.
- Dry-run diff previews changes before writing.
- Data moves directly from warehouse to destination.
- Integrates with Dagster, Prefect, Airflow, and GitHub Actions.
- Apache 2.0 licensed.
- Managed cloud hosting is outside drt-core.
- RBAC and audit logging are outside drt-core.
- Syncs run in batches, not reverse syncs.
Plans
Every plan and pricedrt-core
Free plan
The verdict
drt gives repository-oriented teams a self-hosted way to move modeled warehouse data into team tools. Teams needing RBAC, audit logging, plugins, or managed cloud hosting should note those are outside drt-core.
What the maker publishes
- AI integration
- An optional MCP server lets compatible clients list syncs, run syncs, validate configuration, and inspect status. source
- CI/CD
- The project provides a GitHub Action for scheduled or event-triggered syncs and documents structured status outputs. source
- Connectors
- The project lists 13 sources and 34 destinations, including BigQuery, Snowflake, Postgres, Slack, and Salesforce. source
- Data handling
- The README says data goes directly from the user's warehouse to the destination and never through a drt-hosted intermediary. source
- dbt lineage
- drt can read dbt projects and use `ref()` models so activation follows the model lineage. source
- Enterprise boundary
- The open-core document places RBAC, audit logging, plugin system, and managed cloud hosting outside drt-core. source
- Intended users
- The site describes drt as suited to teams that ship with dbt and Git and want reverse ETL pipelines managed in their repository. source
- License
- The project is licensed under Apache 2.0. source
- Orchestration
- The README lists integrations for Dagster, Prefect, Airflow, and a dbt manifest reader. source
- Preview
- `drt run --dry-run --diff` previews which rows will be added, updated, or deleted before writing. source
- Purpose
- drt moves modeled warehouse data into team tools as reverse ETL, with syncs defined in YAML and versioned in Git. source
- Security
- The security policy describes Dependabot, CodeQL, pip-audit, and CycloneDX SBOM generation for releases. source
- Support
- The security policy asks vulnerability reports to be emailed and says receipt is acknowledged within 48 hours, with critical fixes or mitigations within 7 days. source
- Telemetry
- Telemetry is off by default; when enabled, the project says it collects sync metadata and does not transmit sync names, SQL, destination URLs, credentials, or project paths. source
- Workflow
- Users install `drt-core` with pip and run syncs from the terminal using `drt run`. source
Checked against the maker's pages on 8 Oct 2026.