Mage is an Apache-2.0 licensed, self-hosted development environment for building, running, and managing data pipelines, aimed at data engineers, data scientists, and analytics teams who want to develop ETL, ELT, and orchestration workflows locally before scaling them.

What it is

Mage OSS is a Python project that provides a self-hosted development environment for creating production-grade data pipelines. It lives in the data engineering and orchestration ecosystem, and it combines a modular, notebook-style editor with execution, scheduling, and debugging so that pipeline logic can be written, documented, and run from one place. Pipelines are assembled block by block using Python, SQL, or R, and the same interface is used to trigger runs manually or on a schedule with cron support.

The concrete problem it solves is the gap between local pipeline development and a full orchestration platform. Instead of standing up a cloud service before writing a single transform, a developer can install Mage OSS and work entirely on their own machine, without a cloud account, using prebuilt connectors to databases, APIs, and cloud storage. What it replaces is the ad hoc combination of scripts, notebooks, and scheduler configuration that teams otherwise maintain by hand; Mage folds those pieces into one environment, and when the work outgrows local development, the documented path forward is Mage Pro.

Key capabilities

• Modular pipelines built block-by-block in Python, SQL, or R through a notebook-style user interface.
• Interactive notebook editor for writing and documenting transformation logic alongside the pipeline definition.
• Prebuilt connectors for databases, APIs, and cloud storage, so integrations do not require custom client code.
• Scheduling that runs jobs manually or on a cron schedule.
• Visual debugging with step-by-step execution, logs, data previews, and error handling.
• dbt support, allowing dbt models to be built and run directly inside Mage.
• Three local install paths: the mageai/mageai:latest Docker image, pip install mage-ai, and conda install -c conda-forge mage-ai.

Who uses it and how

• Data engineers moving data from Google Sheets to Snowflake with a Python transform.
• Analytics teams scheduling a daily SQL pipeline to clean and aggregate product data.
• Analytics engineers developing dbt models inside a visual, notebook-style interface rather than a terminal alone.
• Developers running simple ETL and ELT jobs locally with full transparency, without provisioning a cloud account.
• Teams that start on a local machine and later move to Mage Pro when they need multi-environment orchestration, role-based access control, real-time monitoring and alerts, and CI/CD.

Getting started

Docker is the recommended install: docker pull mageai/mageai:latest. Alternatives are pip install mage-ai and conda install -c conda-forge mage-ai, with the full setup guide at docs.mage.ai.

How it compares

Among the tools named in the project's own facts, dbt is the closest relative: Mage supports building and running dbt models inside its interface, so it can serve as the orchestration and development layer around dbt rather than a replacement for model authoring. The sharper dividing line is internal, between Mage OSS and Mage Pro — Mage OSS is Apache-2.0, self-hosted, and requires no cloud account, while Mage Pro is the commercial platform adding AI-assisted development, multi-environment orchestration, role-based access control, monitoring and alerts, CI/CD, and version control, available fully managed, hybrid, or on-premises.

When to use it — and when not to

Choose Mage OSS when the goal is developing and running pipelines on a local machine with an Apache-2.0 licence and no cloud dependency. A self-hoster must operate the environment and its execution themselves, and the production-grade features teams often need — role-based access control, real-time monitoring and alerts, CI/CD, multi-environment orchestration — sit in the paid Mage Pro tier rather than in this repository. Anyone needing managed infrastructure, enterprise governance, or vendor support from day one should look at Mage Pro or a comparable hosted platform instead, and the 624 open issues alongside a README that reads more as product marketing than architecture documentation are worth weighing before committing.

docker pull mageai/mageai:latest

pip install mage-ai

conda install -c conda-forge mage-ai