migration-check

name	migration-check
description	Verify database schema matches code queries before deploying. Catches missing columns, broken joins, schema drift. Use before every deploy that touches DB queries.
labels	["deploy","migration","db","verify"]
keywords	["migration","schema","column","database","supabase","deploy check"]
entrypoint
fuel_grade	diesel
trust_tier	4
version	1.0.0

Migration Check

Before deploying code that changes database queries, verify the schema exists.

What to check

1. Every .select('...') in Supabase queries — do those columns exist?
2. Every .from('table') — does the table exist?
3. Every join (.select('..., relation:other_table(...)')) — does the FK relationship exist?
4. Any new columns referenced in code — is there a migration for them?

How to check (Supabase)

# List columns in a table
curl -s "${SUPABASE_URL}/rest/v1/dnu_cities?select=*&limit=0" \
  -H "apikey: ${SUPABASE_KEY}" -I | grep -i "content-profile"

# Or query information_schema
# SELECT column_name FROM information_schema.columns WHERE table_name = 'dnu_cities';

Rule

If code references a column that doesn't exist in the database, the deploy MUST include a migration that adds it. No exceptions.

Silent return [] on Supabase errors hides these failures. The cities page was empty for this reason on 2026-04-05.

Prevention

Before merging any PR that modifies a .select() or .from() call:

1. Check if new columns/tables exist in production DB
2. If not, create and run migration first
3. Run post-deploy check after deploy