Documentation
Install Brightfield, render your first crosswalk, then go deeper: the open datasets, and Finetype, Arcform and Dovetail, the tools that build them.
You've opened a file and want to know what's in it before you trust it. These docs show you how to draw it with Brightfield, query Meridian's open datasets in your browser or anywhere else, and type and check your own columns with the tools that build those datasets.
Start with Brightfield. One command, one binary, with no server, no webview and no account:
curl -fsSL https://install.meridian.online/brightfield | bashPrefer Homebrew? brew install meridian-online/tap/brightfield.
That installs Brightfield, Meridian's native macOS app: chart specs and protocol graphs, rendered on your GPU. It's early access — v0.1.8, macOS only, Apple silicon and Intel — and the docs are plain about what works today and what doesn't yet.
Your first five minutes
Run brightfield and the window opens on a gallery of starting points that
ship inside the binary. Open the EDGAR ↔ GLEIF crosswalk and a real ten-source
protocol draws as a lineage graph — then come back Home whenever you like. The
quick start walks it step by step.
No Mac, or nothing to install? The same crosswalk is queryable in your browser at /datasets/library/edgar_lei.
What is Meridian?
Meridian is open-source software for people who open a file and need to trust what is in it. It has three parts: Brightfield, a native desktop app in early access; selected public datasets, maintained by Meridian, that you can search and query in the browser; and Finetype, Arcform and Dovetail, the tools that build those datasets.
Brightfield
The native macOS app — chart specs and protocol graphs rendered on your GPU. Early access, honestly labelled.
Datasets
Open reference data — browse it in the explorer, query it with SQL, take it anywhere with one ATTACH line.
The toolchain
The datasets and the tools are one project: type every column, verify every relationship, make every pipeline reproducible.
Finetype
Semantic type inference for messy data — 251 types across 65+ locales, pure Rust, with a DuckDB extension.
Dovetail
Discovers how unfamiliar data loads and how tables relate — then compiles it to runnable SQL.
Arcform
Local-first, asset-aware data pipelines: YAML steps, DuckDB engine, one Rust binary.