Seeding Data

Zeroback doesn’t have a dedicated seed command — instead, you write a regular mutation and run it with the CLI. This keeps seeding type-safe, transactional, and composable with the rest of your code.

1. Create a seed file

Add a file in your zeroback/ directory:

zeroback/seed.ts

import { mutation } from "./_generated/server"

export const run = mutation(async ({ db }) => {
  // Guard against re-seeding
  const existing = await db.query("tasks").first()
  if (existing) {
    console.log("Already seeded, skipping.")
    return "already seeded"
  }

  await db.insert("tasks", { text: "Buy groceries", completed: false })
  await db.insert("tasks", { text: "Walk the dog", completed: false })
  await db.insert("tasks", { text: "Read a book", completed: true })

  return "seeded 3 tasks"
})

2. Run it

Make sure the dev server is running:

zeroback dev

Then in another terminal:

zeroback run seed:run

The function name follows the pattern filename:exportName.

Passing arguments

You can pass JSON arguments to your seed function:

zeroback/seed.ts

import { mutation } from "./_generated/server"

export const run = mutation(async ({ db }, { count }: { count: number }) => {
  for (let i = 0; i < count; i++) {
    await db.insert("tasks", {
      text: `Task ${i + 1}`,
      completed: false,
    })
  }
  return `seeded ${count} tasks`
})

zeroback run seed:run '{"count": 10}'

Seeding a deployed instance

By default, zeroback run targets http://localhost:8788. To seed a deployed instance, pass its URL:

zeroback run seed:run --url https://your-app.your-subdomain.workers.dev

Tips

• Always guard against re-seeding so the script is safe to run multiple times.
• Seed functions are regular mutations — they’re transactional and type-safe.
• You can create multiple exports for different scenarios (e.g. seed:users, seed:products).
• For large datasets, split inserts across multiple mutations to avoid transaction size limits.