Show HN: Graphene – Data analysis toolkit for your coding agent

Two developers who previously worked at several business-intelligence companies released Graphene, an open-source data analysis toolkit that gives coding agents a semantic layer, MDX-like dashboard files, and connectors to popular data warehouses or local DuckDB. Graphene's semantic layer is designed to be more token-efficient than YAML and more deterministic than Markdown, with composable, callable metric macros and a SQL query API, and the project is available on GitHub at graphene-data/graphene with an example project at graphene-data/example-flights. The developers said the toolkit lets an agent add instrumentation, adjust pipelines and transformations, and add dashboards in a single pull request.

My friend and I have worked at a number of BI companies and thought: Can’t a coding agent do most of this now? Almost. They just need a context/semantic layer to ensure query correctness and some kind of artifact for publishing findings and visualizations. We built an open source project called Graphene that provides these tools: - Semantic layer that’s more token-efficient than YAML, more deterministic than Markdown eg, composable, callable metric macros , with a query API that’s well in-distribution SQL . - MDX-like files for dashboards Markdown with inlined SQL + HTML components for viz with support for CSS and Javascript - Connects to popular data warehouses or local DuckDB Your coding agent will build really in-depth reports and can ofc leverage any other skills or context that you've made available to it. We dogfood it inside a monorepo with our website, app source code, planning docs, etc so the agent has access to a ton of context. It’s also nice that the agent can add instrumentation, adjust pipelines and transformations, and add dashboards all in one PR. If you want to try it out, just point your coding agent at https://github.com/graphene-data/graphene/blob/main/docs/set... https://github.com/graphene-data/graphene/blob/main/docs/setup.md and ask it to set up Graphene. If you don’t have data to play with, you can clone our example project: 1. git clone < https://github.com/graphene-data/example-flights.git https://github.com/graphene-data/example-flights.git 2. cd example-flights && npm install Would love any/all feedback Comments URL: https://news.ycombinator.com/item?id=49927295 https://news.ycombinator.com/item?id=49927295 Points: 2 Comments: 0