# Mock data bundle — D2C E-commerce Brand

This bundle is synthetic data used to power the multi-agent operating
model dashboard described in the case study at
`/casestudies/ecommerce/`. Use it to scaffold your build without needing
production access.

> **Important:** This is mock/synthetic data, not real customer data.
> The numbers and names are random. Replace with your own pipeline
> before going live.

## Files

| File | Rows | Purpose |
|------|------|---------|
| `entities.csv` | 60 | The skus the 5 agents (DMAA, IFRA, CSWA, PFAA, SKPA) operate over. |
| `daily_signals.csv` | 300 | Last 30 days of per-sku operational signals (sample of 10 entities). |
| `decisions_history.jsonl` | 150 | 30 days of Daily Decision Lists the CSOA produced. |
| `agent_writebacks.jsonl` | 60 | Recent forecasts/decisions written by each of the 5 agents. |
| `outcomes.csv` | 50 | Approved-decision outcomes for nightly CSOA retraining. |
| `playbooks.md` | n/a | Stubs for 3 D2C E-commerce Brand-specific playbooks: Sale-event surge playbook, Fraud-block playbook, SKU-delisting diligence. |
| `regulatory_updates.jsonl` | 8 | Recent D2C E-commerce Brand regulatory updates (PDPA, KPDN, JAKIM, MOH). |

## How the prompts use this

- **Build-It Step 4 (Daily Decision List)** — open `decisions_history.jsonl` and `outcomes.csv` to see the schema you should generate.
- **Knowledge Graph Step 1 (Obsidian vault)** — entries in `entities.csv` become individual entity-profile notes in your vault.
- **Knowledge Graph Step 3 (Embedding pipeline)** — feed `playbooks.md` and `regulatory_updates.jsonl` to the ingestion pipeline as your first test corpus.
- **Knowledge Graph Step 5 (Nightly sync)** — `outcomes.csv` is the input that closes the learning loop.

## Quick download

A single ZIP of all files is available at `sample-bundle.zip` next to this README.