Overview
Graphene is a data analytics framework designed around coding agents as the primary user. It supports SQL-based data exploration, visualization, and reporting through an everything-as-code workflow.
Its goal is to help agents produce accurate queries and polished analytics while keeping business logic in a version-controlled repository.
Key Features
- Graphene SQL combines familiar SQL with governed measures and modeled joins.
- Dashboard files support visualizations, filtering inputs, dynamic behavior, and layouts for monitoring dashboards or narrative notebooks.
- The command-line interface gives coding agents a direct way to control projects, while documentation is packaged inside an agent skill.
- ANSI-style SQL support includes CTEs, subqueries, window functions, set operators, and more than 170 functions.
- Pages can express custom presentation and interactivity with HTML, CSS, JavaScript, and ECharts.
Data Sources and Workflow
Graphene currently works with Snowflake, BigQuery, ClickHouse, Postgres, MotherDuck, and local data through DuckDB. After project setup, analytics code can be developed locally and managed alongside the rest of a codebase.
Version control and CI make it possible to review changes, revert agent mistakes, test important dashboards, and handle larger refactors across analytics logic. Agents can validate their work, inspect dashboards, and iterate within a local feedback loop.
Availability
Graphene is presented as free to use. Because analytics definitions and business logic remain in the user's own repository, projects are not tied to a proprietary contract for access to that logic.