# Sync records

A sync reads website data on a schedule and publishes records your app can consume.

`integrations/atlas-demo/syncs/bookings.ts`:

```ts
import { randomUUID } from "node:crypto";
import { createSync } from "@browserlayer/sdk";
import { z } from "zod";
import listBookings from "../connector/flows/list-bookings.js";
import { publicBookings } from "../models/bookings.js";
import { Booking } from "../models/booking.js";
export default createSync({
  description:
    "Keep the application's bookings current, including cancellations and deletions.",
  connection: "required",
  schedule: "every 30 seconds",
  model: Booking,
  checkpoint: z.null(),
  async exec(ctx) {
    const scan = await ctx.step("scan", async () => randomUUID());
    const source = await ctx.run(listBookings, {});
    const records = await publicBookings(ctx, source);
    await ctx.publish("snapshot", records, {
      checkpoint: null,
      scan,
      completeScan: true,
    });
  },
});
```

This example publishes a complete snapshot. The connector and model are supplied by the [Atlas tutorial](/docs/atlas).

## Inspect and control locally

```sh
browserlayer syncs list <connection-id> --local
browserlayer syncs trigger <connection-id> bookings --local
browserlayer syncs records <connection-id> bookings --local
browserlayer syncs pause <connection-id> bookings --local
browserlayer syncs start <connection-id> bookings --local
```

Keep `browserlayer start` running. Connected accounts get schedules automatically. Waiting runs prevent overlap. `pause` prevents the next occurrence; it does not cancel the current run. `trigger` enables the schedule and makes it due now.

Use the same commands with `--env development` after deploying.

| Option       | Required | Example                                |
| ------------ | -------- | -------------------------------------- |
| `connection` | Yes      | `"required"`                           |
| `schedule`   | Yes      | `"every 5 minutes"`                    |
| `model`      | Yes      | Zod object with string `id`            |
| `checkpoint` | Yes      | `z.null()` or a source progress schema |

**Next:** [Checkpoints and deletions](/docs/checkpoints), then [consume records](/docs/records).