Database documentation tool

ErdDocs is a database documentation tool that runs in the browser. Paste a schema dump and you get a data dictionary of every table, column, type, default, key and description, plus an interactive ER diagram. Both print to PDF. There is no install and no account, and the schema is parsed in your browser and never uploaded.

Your schema

CREATE TABLE warehouses (
    id    bigserial   PRIMARY KEY,
    code  varchar(10) NOT NULL UNIQUE,
    city  varchar(80) NOT NULL
);
COMMENT ON TABLE warehouses IS 'Physical locations; code appears on labels';

CREATE TABLE items (
    id       bigserial    PRIMARY KEY,
    sku      varchar(40)  NOT NULL UNIQUE,
    title    varchar(200) NOT NULL,
    weight_g int
);
COMMENT ON COLUMN items.weight_g IS 'Grams; NULL for services and digital goods';

CREATE TABLE stock (
    warehouse_id bigint NOT NULL REFERENCES warehouses(id),
    item_id      bigint NOT NULL REFERENCES items(id),
    on_hand      int    NOT NULL DEFAULT 0,
    reserved     int    NOT NULL DEFAULT 0,
    PRIMARY KEY (warehouse_id, item_id)
);
COMMENT ON TABLE stock IS 'Current quantities; history lives in stock_moves';

CREATE TABLE stock_moves (
    id           bigserial   PRIMARY KEY,
    item_id      bigint      NOT NULL REFERENCES items(id),
    from_wh      bigint      REFERENCES warehouses(id),
    to_wh        bigint      REFERENCES warehouses(id),
    qty          int         NOT NULL,
    moved_at     timestamptz NOT NULL DEFAULT now()
);
COMMENT ON COLUMN stock_moves.from_wh IS 'NULL for inbound deliveries';
COMMENT ON COLUMN stock_moves.to_wh IS 'NULL for write-offs and shipments';

CREATE TABLE warehouses (
    id    bigserial   PRIMARY KEY,
    code  varchar(10) NOT NULL UNIQUE,
    city  varchar(80) NOT NULL
);
COMMENT ON TABLE warehouses IS 'Physical locations; code appears on labels';

CREATE TABLE items (
    id       bigserial    PRIMARY KEY,
    sku      varchar(40)  NOT NULL UNIQUE,
    title    varchar(200) NOT NULL,
    weight_g int
);
COMMENT ON COLUMN items.weight_g IS 'Grams; NULL for services and digital goods';

CREATE TABLE stock (
    warehouse_id bigint NOT NULL REFERENCES warehouses(id),
    item_id      bigint NOT NULL REFERENCES items(id),
    on_hand      int    NOT NULL DEFAULT 0,
    reserved     int    NOT NULL DEFAULT 0,
    PRIMARY KEY (warehouse_id, item_id)
);
COMMENT ON TABLE stock IS 'Current quantities; history lives in stock_moves';

CREATE TABLE stock_moves (
    id           bigserial   PRIMARY KEY,
    item_id      bigint      NOT NULL REFERENCES items(id),
    from_wh      bigint      REFERENCES warehouses(id),
    to_wh        bigint      REFERENCES warehouses(id),
    qty          int         NOT NULL,
    moved_at     timestamptz NOT NULL DEFAULT now()
);
COMMENT ON COLUMN stock_moves.from_wh IS 'NULL for inbound deliveries';
COMMENT ON COLUMN stock_moves.to_wh IS 'NULL for write-offs and shipments';

Your diagram appears here.

A warehouse schema. The comments that explain NULL become the Description column, the part hand-written docs always lose first.

Database documentation tools compared

Most database documentation tools connect to the database. With ErdDocs you bring the schema to the tool. The table compares what ErdDocs does.

Feature	ErdDocs	SchemaSpy	tbls	dbdocs	SchemaCrawler	DBeaver	Atlas	Redgate SQL Doc	Dataedo
Nothing to install: runs in a browser tab	✓ Yes	× No	× No	✓ Yes	× No	× No	× No	× No	Partly: a hosted option through sales
Reads a schema file, no live database needed	✓ Yes	× No	Partly: its own JSON output only	✓ Yes	Partly: a snapshot made from a database once	× No	Partly: runs the files on a dev database	× No	Partly: SQL dump import, in beta
Reads Prisma and ORM models as text	✓ Yes	× No	× No	× No	Not stated	× No	Partly: runs an ORM program	× No	× No
Data dictionary as a printable PDF	✓ Yes	Partly: HTML site	Partly: Markdown and Excel	Partly: web document	Partly: HTML, text, JSON	× No: an open feature request	Not stated	✓ Yes	✓ Yes
The schema stays on your machine	✓ Yes	✓ Yes	✓ Yes	× No	✓ Yes	✓ Yes	Partly: docs on Atlas Cloud	✓ Yes	Partly: in its repository database
No subscription needed	✓ Yes	✓ Yes	✓ Yes	✓ Yes	✓ Yes	Partly: community edition	Partly: docs are on the paid plan	× No	× No

Product names belong to their owners. The table reflects each vendor’s own pages as of November 2026. “Not stated” means the vendor does not say.

SchemaSpy, tbls, SchemaCrawler and DBeaver are installed on your machine and connect to a live database. SchemaSpy, tbls and DBeaver read no ORM models, and none of the four prints a data dictionary to PDF: SchemaSpy writes an HTML site, tbls writes Markdown and Excel, SchemaCrawler writes text, HTML or JSON, and a data dictionary in DBeaver is still an open feature request.

dbdocs builds a web document from DBML in the dbdiagram editor and publishes it on its own site. It reads only DBML and makes no PDF of the document.

Atlas runs your SQL files or ORM program against a dev database, and its schema docs and ERD live on Atlas Cloud on a paid plan.

Redgate SQL Doc and Dataedo do print a document, but both are sold by the year and neither reads ORM models. Redgate SQL Doc is a Windows install that works with SQL Server only. Dataedo keeps the schema in its own repository database, which you install and host unless you arrange hosting with its sales team.

FAQ

What is the fastest way to document an existing database?

Export the schema with pg_dump --schema-only or mysqldump --no-data, then paste it into ErdDocs. The dictionary appears as you paste, and it prints straight to PDF. There is nothing to install, no account and no connection to the database. The whole path takes a few minutes, and most of that is spent reading your own schema.

What are the open source options?

SchemaSpy is a Java tool that connects to a live database over JDBC and generates an HTML site. tbls is a Go command-line tool that connects to a live database and writes Markdown. SchemaCrawler is a Java command-line tool over JDBC. All three need an installed runtime or binary and a connection to the database. ErdDocs needs neither. You paste the schema in a browser tab, and the dictionary and the diagram are there.

What if the documentation is a one-off deliverable?

That is the case ErdDocs is built for: an audit request, a handover, an onboarding doc. You paste the schema, read the dictionary and print the PDF, with nothing to install, nothing to maintain and nothing pointed at production.

Why does privacy matter for a documentation tool specifically?

Because the input is your entire database structure: table names, columns, relations. That is exactly what an attacker wants for reconnaissance and what NDAs often cover. ErdDocs parses the schema inside your browser and never transmits it. There is no server that could log, cache or leak it, and nothing ever connects to the database itself.

What does it cost?

Free without an account: diagrams up to 25 tables, dictionary preview, all image and text exports. One $9 payment unlocks full dictionaries and diagrams for any schema size. It is one payment, not a subscription.

Data dictionary generator: the document itself, in depth.
Dataedo alternative: the enterprise-platform comparison in detail.
Lucidchart alternative: when the job is the diagram, not the drawing.
MySQL ·PostgreSQL: dialect-specific pages.