Skip to content

Developer Guide

This section covers the technical internals of Mirabelle for developers who need to extend, debug, or operate the system.


In This Section

Page Contents
Ports & Services Every service, port, and how to check/restart them
Database Schema Every table in every domain, with columns, types, and relationships
Multi-Tenancy How the master DB + tenant DB model works end to end
Authentication & Security JWT, OAuth providers, token lifecycle, role guards
Allocation Engine Internals Score-based allocation algorithm, 8-pass engine, CH result schema
Replay Engine Internals Iterative supply+allocation replay, planning_date override, stock evolution, API endpoints
Firm Supply Orders — Internals ERP-originated firm orders: loading, merge with planned orders, FEFO simulation
Backend Code Walkthrough FastAPI setup, dependency injection, endpoint patterns, process logging
Frontend Code Walkthrough React 18 SPA, state management, component hierarchy
Data Pipeline Code Step-by-step pipeline walkthrough: ETL → Forecast → MEIO → Supply
Causal SQL Engine SQL-first causal demand engine: CTE chain, failure rate waterfall, deployment filtering, override handling, parity testing
Historization — Archive Date Model Freeze, archive_date, pipeline_version_index, CH partitioning
Planning Date System Virtual "today" per pipeline, resolution chain, business logic rules, CH DEFAULT caveat
Segment & Parameter SQL Tests SQL verification for segmentation membership, parameter resolution, and MEIO multi-parameter assignment

Before you dive in

Read Architecture first for the big-picture data flow. The developer guide assumes familiarity with the pipeline concept (pipeline_id + scenario_id) documented there.