schema

Database Schema

Inspect, visualize, and document the database schema.

Load the database-designer skill for schema documentation patterns.

Usage

/schema [table_name] [--format diagram|table|markdown]

Arguments

Parse $ARGUMENTS to extract:

• table_name (optional) — Focus on a specific table and its relationships
• --format (optional, default: table) — Output format
  • table — Summary tables of columns, types, constraints, indexes
  • diagram — ASCII entity-relationship diagram
  • markdown — Full documentation suitable for a docs/ directory

Instructions

1. Determine Database Type

Check the project for database indicators:

ls prisma/schema.prisma 2>/dev/null         # Prisma
ls knexfile.* 2>/dev/null                    # Knex
ls ormconfig.* typeorm.config.* 2>/dev/null  # TypeORM
ls alembic.ini 2>/dev/null                   # SQLAlchemy/Alembic
ls config/database.yml 2>/dev/null           # Rails

Also check for DATABASE_URL or MYSQL_* env patterns to determine PostgreSQL vs MySQL.

2. Extract Schema

If an ORM schema file exists (Prisma, TypeORM, etc.), read it directly — it's the source of truth.

If MCP database connection is available, use it to introspect:

• List all tables
• For each table: columns, types, nullable, defaults, constraints
• Foreign key relationships
• Indexes

If neither, search for migration files or SQL schema dumps.

3. Output by Format

table (default):

## Schema: my_database

### users
| Column | Type | Nullable | Default | Key |
|--------|------|----------|---------|-----|
| id | uuid | NO | gen_random_uuid() | PK |
| email | varchar(255) | NO | — | UQ |
| name | varchar(100) | YES | — | |
| created_at | timestamptz | NO | now() | |

Indexes: idx_users_email (email)
Relations: → posts.author_id, → comments.user_id

diagram:

┌─────────┐       ┌─────────┐
│  users  │───┐   │  posts  │
├─────────┤   │   ├─────────┤
│ id (PK) │   └──→│ author_id│
│ email   │       │ id (PK) │
│ name    │       │ title   │
└─────────┘       └─────────┘