I got tired of hand-writing fake data for Prisma, so I built a VS Code extension that does it from your schema

Every new table meant the same ritual: open a seed script, hand-write fake
rows, remember which columns need to stay unique, make sure foreign keys
actually point at something that exists. Small thing, but it adds up every
single time you touch the schema.

So I built SeedForge — a VS Code extension that reads your
schema.prisma, figures out sensible fake values per column (emails look
like emails, booleans respect their defaults, enums actually vary instead of
collapsing to one value), and writes the result straight to a file.

What it does (free)

• Parses your Prisma schema — models, scalars, enums, @default, optional fields, arrays.
• Generates realistic values via @faker-js/faker, with light heuristics based on column names.
• Exports to JSON, SQL INSERT, or CSV.
• Fully local. No account, no telemetry, no network calls.

How to use it

1. Open a folder with a schema.prisma.
2. Cmd/Ctrl+Shift+P → SeedForge: Generate Seed Data.
3. Pick a model, pick a row count, pick a format. Done.

It's free on the VS Code Marketplace:
https://marketplace.visualstudio.com/items?itemName=Avers.seedforge

Would genuinely like feedback — especially from anyone whose schemas have
gnarly relation chains. That's the part I'm working on next (generating
correctly-linked parent/child rows in one pass).