# Import & Export

> Move schemas in and out of Sqemo: SQL DDL in seven dialects, DBML round-trip, portable project files, CSV and Markdown reports, images, and print.

Sqemo is not a silo. Schemas come in from the databases you already have
and go out to the formats your pipeline expects.

## SQL DDL

Import `CREATE TABLE` DDL and export it back in seven dialects:
**MySQL, PostgreSQL, Oracle, SQL Server, SQLite, H2, and CUBRID**.

On import, primary keys, foreign keys (including inline `REFERENCES`
column constraints), and comments are recognized — foreign keys become
relationship lines on the canvas. Import and export operate on the
physical model.

## DBML

DBML imports and exports round-trip, so you can move between Sqemo and
text-first tools like dbdiagram without retyping a schema.

## Project files (`.erd.json`)

The project file carries everything: logical and physical models, the
word list, naming rules, and the schema version. It is a plain JSON file —
commit it to your repo, diff it in review, send it to a colleague. The
[MCP server and CLI](/docs/mcp/) operate on the same format.

## Word lists, domains, and glossary

The standards side moves as CSV, so it survives a trip through a
spreadsheet and back:

- **Word list** — export as CSV or JSON, import either back. This is how a
  team's abbreviations arrive from an existing convention document. The
  Korean government's common standard word list (the data.go.kr CSV, Korean
  headers) imports as-is: abbreviations become physical words and retired
  words are skipped.
- **Domains** — export and import as CSV.
- **Glossary** — export the [term definitions](/docs/business-glossary/) as
  CSV.

Word list and domain imports are disabled while a project is connected to a
team standard, since the standard is the source of truth there — see
[Naming standards](/docs/naming-standards/).

## Specification reports (CSV · Markdown)

Four report exports produce the tables that usually get rebuilt by hand for
design reviews and audits — entity, column, index, and relationship
definitions. Each is a flat CSV of the physical model with the logical
names alongside, so a reviewer who does not open the tool can still read
what the schema says.

For a wiki or a pull request, **Export schema document (Markdown)** puts
the same content in a single `.md` file — a mermaid `erDiagram` block that
GitHub, Notion, and Obsidian render as a picture, an entity index, a
column and index table per entity, and the relationship table. Headers
follow the app's UI language, so the file pastes into a Korean or English
team page as is. GitHub draws the diagram on its own; in Notion, paste the
file into a page and switch the Mermaid code block to its preview mode with
the toggle in the block's top-right corner. The same **scope** choice as
images and print (whole diagram, selected entities, or a subject area) lets
you produce a short document covering just the part under review — picking
a subject area adds its name to the file name and title, and entities
outside the scope, along with relationships that cross it, are left out.

## Images and print

Export the canvas as **PNG or SVG** for design docs and reviews, or use
**PDF / Print** to lay the diagram out on paper — A4 or A3, portrait or
landscape — and print it or save it as a PDF from the browser dialog.

## Migration SQL (Pro)

Beyond exporting the whole schema, Sqemo can generate the `ALTER` script
between a saved baseline and your current model — see
[Connect to a live database](/docs/live-database/) for the related drift
check. This one is a Pro feature.

## From the command line

The CLI (part of the MCP package, no login required) exports a project
file without opening the app:

```bash
npx sqemo-mcp export schema.erd.json --format sql --dialect postgres > schema.sql
npx sqemo-mcp export schema.erd.json --format dbml > schema.dbml
```

---
Source: https://sqemo.com/docs/import-export/ · Updated 2026-09-22