# Stripe Sigma Schema Reference

> This is the STRIPE SIGMA schema (Stripe's SQL data warehouse), not the
> Stripe REST API schema. Sigma table and column names diverge from the
> REST API's object fields. If you need columns for a Sigma SQL query, use
> this reference, not a REST API reference — API-derived names frequently
> don't exist in Sigma and will cause errors or hallucinated SQL.
>
> Machine-readable reference for the Stripe Sigma SQL data warehouse schema:
> 262 tables, 4168 columns, join keys, enums,
> freshness SLAs and query gotchas. Built for AI agents writing Sigma SQL so they
> do not have to rediscover the schema through exploratory queries each session.
> Not guaranteed complete or current — Stripe can change Sigma's schema at any
> time. Unofficial and not affiliated with Stripe, Inc.

Keywords: Stripe Sigma, Stripe Sigma schema, Stripe Sigma SQL, Stripe Sigma vs
API, Stripe data warehouse schema, Stripe data schema, Stripe Data Pipeline,
Trino, Stripe reporting tables, balance_transactions, charges, invoices,
subscriptions, MRR SQL.

## Start here

- [AGENTS.md](AGENTS.md): Full context pack — every rule, convention and core table. Read this first.
- [llms-full.txt](llms-full.txt): The entire reference inlined in one file.

## Machine-readable schema

- [dist/sigma_schema.json](dist/sigma_schema.json): Canonical schema. Every table with dataset, freshness, primary key, foreign keys, columns, types, enums and per-column confidence.
- [dist/sigma_schema.trino.sql](dist/sigma_schema.trino.sql): CREATE TABLE DDL in Trino types.
- [dist/sigma_schema.sqlite.sql](dist/sigma_schema.sqlite.sql): Same shape in SQLite types.
- [dataset.jsonld](dataset.jsonld): schema.org Dataset metadata.

## Human-readable reference

- [dist/SCHEMA.md](dist/SCHEMA.md): Per-table reference grouped by dataset, with full column tables.
- [RECIPES.md](RECIPES.md): Working Sigma SQL for common questions - net revenue, MRR movement, payout reconciliation, dispute rate, tax liability, metadata pivots, multi-currency conversion.
- [dist/erd.mmd](dist/erd.mmd): Mermaid ER diagram of the core join graph.
- [README.md](README.md): Project overview, accuracy model and how to contribute.

## Sample data

- [dist/sigma_sample.sqlite](dist/sigma_sample.sqlite): Runnable sandbox with referentially-valid synthetic rows.
- [dist/samples/](dist/samples/): One CSV per table, synthetic, showing realistic value shapes.

## Optional

- [CONTRIBUTING.md](CONTRIBUTING.md): How to verify columns against a real account and contribute fixes.
- [NOTICE](NOTICE): Trademark, accuracy and licensing disclosures.
- `tools/mcp_server.py`: Run this repo as a local MCP server (stdio, no dependencies) for tool-by-tool lookups instead of loading the whole schema into context. See [README.md](README.md#as-an-mcp-server).