# End-to-end workflow

> A summer flash sale walkthrough showing the Shoplazza AI Toolkit chaining the products, discounts, orders, and shop skills in a single agent session.

A realistic campaign shows how the toolkit strings multiple domains together in one session. You describe the goal; the agent picks the right skill for each step and follows the safety protocol on every write.

## Scenario: run a summer flash sale

### 1. Add the product

```
/shoplazza-products Add a new product: Summer print T-shirt, priced 129, main image from https://cdn.example.com/tee.png
```

The agent creates the product and captures its ID.

### 2. Put it in a collection

```
/shoplazza-products Publish product 445566 and add it to collection 8899
```

### 3. Launch the flash sale

```
/shoplazza-discounts Run a limited-time flash sale on the products in collection 8899
```

Creates an auto-applied flash sale scoped to that collection — no coupon code needed.

### 4. Announce it

```
/shoplazza-shop Publish a blog post "Summer Flash Sale is live" to blog 123
```

### 5. Fulfill orders as they land

```
/shoplazza-orders Ship order 660011 with tracking SF1234567890
```

### 6. Handle a return safely

```
/shoplazza-orders Refund 15.50 to the buyer on order 660011, note "wrong size", restock the item
```

Because this moves money, the agent previews with `--dry-run`, restates the amount and restock, and waits for your confirmation.

### 7. Review the results

```
/shoplazza-shop Show me the top-selling SKUs for the sale period
```

Runs a `shop analytics by-sku` query over the campaign window.

## Why it works

Each step returns structured JSON, so the agent carries the output of one command into the next — creating, scoping, promoting, fulfilling, and reporting — while switching skills as the domain changes and confirming every write. That is the whole toolkit working as one.

## See also

* [Usage examples](./usage/products)
* [Safety & permissions](./safety-and-permissions)