David Toolkit

A browser-first intelligence center for independent shop owners.

David Toolkit starts where most small shops already are: an old POS, exported spreadsheets, supplier deadlines, and too much judgment trapped in one person's head. This repo turns that reality into a clean, open-source intelligence layer.

This first demo is built around a real independent food shop context. It reads retail exports, applies a curated correction layer, overlays weather and calendar context, and renders a founder-grade dashboard that separates raw evidence from interpreted signals.

What Works Today

• Parses ExportStatVente, STA_satvente, STA_ratioCAT, and Analyse_00 style exports
• Builds a browser-ready intelligence center with:
  • supplier command panels
  • product and category intelligence
  • raw vs interpreted demand signals
  • weather and holiday overlays
  • 3-year macro context when a finance workbook is present
• Ships with a sanitized sample dataset so outsiders can run the demo without private shop data
• Includes a verification script that checks parser and scoring assumptions
• Supports a French operator-facing product surface through sample-data/config/context.json while keeping the repo and docs in English

Quick Start

npm install
npm run build:demo   # requires Gold data in data/gold/
npm run serve

Open http://localhost:4173.

Or build everything from raw CSVs:

npm install
npm run build:full   # import → gold → demo in one step
npm run serve

Using Your Own Data

Place your POS CSV exports in data/real/ (gitignored). The pipeline auto-detects 7 file types by header fingerprint: monthly stats, annual stats, transactions, category mix, margin analysis, hourly patterns, and product master.

Then run:

npm run build:full

Repository Guide

• ARCHITECTURE.md explains the importer, scoring model, and front-end structure.
• DATA_SOURCES.md documents every expected file and mapping layer.
• DEMO.md gives a demo script for talks, videos, and a Show HN style launch.
• ROADMAP.md shows what comes next.
• docs/ISSUE_BACKLOG.md seeds the first public issue backlog.
• docs/CURSOR-SETUP.md defines the recommended Cursor, Slack, mobile, and MCP workflow for this repo.
• docs/CURSOR-AUTOMATIONS.md contains the first automation specs worth creating.

Contribution Standard

This repo is meant to be contributor-ready from day one.

• Read CONTRIBUTING.md before opening a pull request.
• Respect the CODE_OF_CONDUCT.md.
• Report security issues through SECURITY.md.
• Agent-based contributors should also read .cursor/BUGBOT.md and the PR template expectations.

Why AGPL

The point of David Toolkit is not just “source visible.” It is a real commons project. If someone runs an improved hosted version, those improvements should stay in the commons. That is why the code is licensed under AGPL-3.0-only.

Status

v0.1.0 — Shipped 2026-03-13 after a 5-day solo hackathon.

Tabs 1–4 are functional with real data. Tabs 5–7 are documented stubs. The pipeline works on the main workstation with POS CSV exports.

→ Live dashboard → Contributor Guide — start here if you want to continue this → Day 4 Devlog — the honest technical brief