KeyanVakil/seed-db

Populate the database with believable, demo-ready data.

Steps

1. Find the database schema (look for schema files, migrations, or ORM models).
2. Understand the relationships between tables — seed in dependency order (parents before children).
3. Generate seed data that looks real:
   • Names: actual human names, not "User 1"
   • Dates: recent and spread out naturally, not all the same timestamp
   • Content: realistic sentences/descriptions, not lorem ipsum
   • Numbers: plausible ranges (ages 18–65, prices $9.99–$299, ratings 3.2–4.9)
4. Write a seed script using the project's existing DB client. Match the style of any existing seed/fixture files.
5. Run the seed script.
6. Verify by querying a count from each seeded table.

Data volume defaults

• Main entity (users, projects, posts): 10–20 records
• Related entities: 2–5 per parent
• Junction/many-to-many: realistic distribution, not every combination

Rules

• Never truncate/drop tables unless the user explicitly says to reset
• If a seed script already exists, add to it rather than replace it
• Use deterministic data (hardcoded arrays) not random — results should be reproducible
• If there's a created_at field, spread dates over the last 90 days
• Prioritize data that makes the UI look good for a demo (variety, completeness, no awkward empty states)