Stacks Documentation
SkillsChangelog
TeamSponsorsPartnersPostcardwareAwesome StacksContributing

Migrations

Copy page
Copy page as Markdown for LLMs
View as Markdown
View this page as plain text
Open in ChatGPT
Copy content and open ChatGPT
Open in Claude
Copy content and open Claude
Open in Perplexity
Search with Perplexity AI
On this page 6

stacks-migrations · Data layer · model-invoked

Migrations here are derived from your models rather than hand-written. You change the model, buddy generate:migrations diffs it against the schema and emits the SQL, and you read that SQL before applying it. This skill covers the generation workflow, the naming conventions and the 96+ migrations that ship.

When to reach for it

  • Creating migration files
  • Running migrations
  • Fresh migration (drop + recreate)
  • Seeding after migration
  • Migration file naming conventions
  • The 96+ built-in migration files

Inside the skill

The sections an agent reads once the skill loads.

  • Key Paths
  • CLI Commands
  • Creating a Migration
  • Migration Generation from Models
  • Built-in Migrations (96+)
  • Workflow
  • Gotchas

Where the code lives

  • Migration files: database/migrations/ (96+ files)
  • Database config: config/database.ts
  • Model snapshot: storage/framework/database/model-snapshot.<dialect>.json

Using it

This one is model-invoked. Your agent reaches for it on its own when the task matches, and you can also call it by name:

/stacks-migrations

Source: stacks-migrations/SKILL.md. Shadow it for one project with app/Skills/stacks-migrations/SKILL.md, then re-run buddy setup:ai. See Writing your own.

Native source verification

The skill has been checked against current source and retained tests. Read its entrypoint and linked references for the actual contract; summaries here are discovery pointers. Native capabilities maps every core package and records driver evidence without claiming universal parity.

migrate:align --relax-constraints also plans removal of legacy NOT NULL constraints where the current model allows null, and foreign keys on polymorphic id/type pairs where the model declares no single-table reference. It never makes nullable data required or relaxes primary keys. Inspect the plan, then add --apply to run it. The default continues to report these differences without changing them.