AI Integration (MCP)

Updated August 28, 2026

sqemo-mcp is a local MCP (Model Context Protocol) server that lets AI agents work with your ERDs: read and edit entities, attributes, and relationships; import and export SQL and DBML; generate and check physical names against your team standard; lint, auto-layout, and diff; and, on Pro, read a live database to import it or check it for drift. It runs via npx — no separate install.

Requirements

Node.js 22 or later.

Client setup

Claude Code — add to .mcp.json at your project root:

{
  "mcpServers": {
    "sqemo": { "command": "npx", "args": ["-y", "sqemo-mcp"] }
  }
}

Claude Desktop — add the same mcpServers entry to the config file:

• Windows: %APPDATA%\Claude\claude_desktop_config.json
• macOS: ~/Library/Application Support/Claude/claude_desktop_config.json

Codex CLI — register with one command:

codex mcp add sqemo -- npx -y sqemo-mcp

Or add the equivalent TOML to ~/.codex/config.toml:

[mcp_servers.sqemo]
command = "npx"
args = ["-y", "sqemo-mcp"]

Most other MCP clients (Cursor, etc.) use the same mcpServers JSON.

Using an AI agent already? Just paste this page’s URL into Claude Code, Codex, or Cursor and ask it to connect the sqemo MCP server — the steps above are everything it needs.

Sign in for server ERDs

No login is needed for local .erd.json files — every file tool works offline. To read or edit ERDs saved on the Sqemo server, sign in once from a terminal:

npx sqemo-mcp login    # pick Google, GitHub, or email + password
npx sqemo-mcp logout   # remove stored credentials

login asks how you want to sign in. Google and GitHub open a browser tab and hand the result back to the terminal; the third option takes an email and password without leaving it. Whichever you pick, only a refresh token is stored, in ~/.erdmaker/credentials.json — never your password.

No browser on that machine? Use npx sqemo-mcp login --password, or set SQEMO_EMAIL / SQEMO_PASSWORD for unattended runs.

CI without an agent

The package doubles as an offline CLI for pipelines:

# fail the build on naming-standard violations (exit code 1)
npx sqemo-mcp lint schema.erd.json

# export to SQL or DBML on stdout
npx sqemo-mcp export schema.erd.json --format sql --dialect postgres > schema.sql

GitHub Actions:

- run: npx sqemo-mcp lint schema.erd.json

CI drift check (Pro)

On Pro, lint can also compare your ERD against a real database and fail CI when they’ve drifted apart. For a full step-by-step walkthrough — including setting a password if you signed up with Google/GitHub, and trying it with Docker if you don’t have a database handy — see Connect to a Live Database.

# .github/workflows/schema-drift.yml
- run: npx sqemo-mcp lint erd/app.erd.json --db "$DATABASE_URL" --ignore "flyway_*"
  env:
    DATABASE_URL: ${{ secrets.DATABASE_URL }}   # read-only account recommended
    SQEMO_EMAIL: ${{ secrets.SQEMO_EMAIL }}     # Sqemo Pro account (password sign-in;
    SQEMO_PASSWORD: ${{ secrets.SQEMO_PASSWORD }} # OAuth-only accounts: set a password first)

SQEMO_EMAIL/SQEMO_PASSWORD sign in for the duration of the CI job — no login step or stored token needed. Interactive sign-in can use your browser, but a CI job has no browser, so it needs a password. If your Sqemo account was created via Google/GitHub OAuth it has none yet — set one once, then use it here. Introspection itself runs inside your CI job and only queries information_schema — no database credentials are ever sent to Sqemo.

Prefer zero database credentials in CI? Compare against a schema dump instead, with no login required either:

npx sqemo-mcp lint erd/app.erd.json --schema schema.sql --dialect postgres

What agents can do

39 tools in total. A few of them: get_erd_overview, upsert_entity, upsert_attribute, upsert_relationship, upsert_attributes, upsert_dictionary_words and upsert_domains (bulk edits in one save), import_sql, import_dbml, export_sql, export_dbml, generate_physical_name, check_naming, lint_erd, auto_layout, diff_erds, proposal tools for suggesting new words to your team standard (a person still approves or rejects them — that side is a two-minute video), and two Pro tools for working against a real database: introspect_db (import a live PostgreSQL/MySQL schema into an ERD, read-only) and check_db_drift (check whether a live database or schema dump has drifted from the ERD).