Sqemo is an ERD tool built around one idea: physical names should be generated from your team's naming standard, not typed by hand.

You register a glossary (customer → cust, number → no) and naming rules (case style, delimiter, how to handle unknown words) first. Then you model at the logical level — "Customer Number" — and the physical column comes out cust_no, the same way for everyone, every time. Overrides are allowed but explicitly flagged, so they never spread silently. Because the correct name is computable, drift is detectable: an in-app linter flags deviations, and a CLI fails CI on violations (npx sqemo-mcp lint).

Key capabilities: • Logical/physical model separation with stable links between the two • SQL DDL import/export in seven dialects: MySQL, PostgreSQL, Oracle, SQL Server, SQLite, H2, CUBRID • DBML import/export — one round trip to move from dbdiagram • Runs free in the browser with no sign-up; projects autosave locally and live in a plain .erd.json file you can commit • MCP server (sqemo-mcp) with 33 tools — AI agents read and edit your ERD while following your naming standard • Team workspaces: shared standard, propose→approve queue for new glossary words, and a compliance view showing which ERDs drift from the standard