Introducing PgDesigner — A Modern Visual PostgreSQL Schema Designer
Today we’re releasing PgDesigner — a visual PostgreSQL schema designer built for teams that treat their database schema as seriously as their application code.
Why we built PgDesigner
Every PostgreSQL team faces the same challenges:
- No single source of truth — the schema lives in migrations, developer heads, or an outdated wiki diagram
- Migrations are scary — writing ALTER statements by hand means risking data loss, table rewrites, or exclusive locks in production
- Existing tools are either generic or outdated — DrawDB supports 4 databases but none deeply. pgModeler is powerful but 19 years old with poor git integration
PgDesigner addresses all three: a visual ERD editor that generates production-ready SQL, compares schemas with hazard detection, and validates with 75 lint rules.
What makes PgDesigner different
We’re the only tool that combines all four:
- PG-specialized — built exclusively for PostgreSQL 18. 70+ data types, partitions, temporal constraints, virtual generated columns, IDENTITY — all first-class
- Git-friendly format — .pgd XML produces single-line diffs when you add a column. Split-model for 50+ table schemas
- Diff/ALTER engine — two models in, safe migration out. 6 hazard codes warn about data loss, table rewrites, and exclusive locks
- Modern visual UI — browser-based ERD canvas with auto-layout, minimap, and dark/light theme
No other tool has all four. pgModeler has 3/4 (no hazard detection, poor git diffs). DrawDB has 1/4 (visual only). Atlas has 2/4 (no visual editor, multi-database abstraction).
Key features at a glance
Visual ERD Canvas — drag tables, zoom, auto-arrange. Tested on 120+ tables. Click to edit columns, constraints, indexes. Create FK by clicking two tables.
DDL Generation — 15-phase topological sort. FK always after tables. Materialized views sorted by inter-dependencies. 99.7% round-trip fidelity on 630+ tables across 6 real databases.
Diff & ALTER Engine — compare old.pgd vs new.pgd → safe ALTER SQL. Hazard detection: DELETES_DATA, TABLE_REWRITE, TABLE_RECREATE, BACKFILL_REQUIRED, DETACH_PARTITION, REATTACH_PARTITION. Compatible cast detection (varchar widening, integer → bigint).
75 Lint Rules — 32 errors (structural integrity), 21 warnings (best practices), 13 info (antipatterns). 15 rules with one-click autofix. Missing FK index? Click fix. char(n) column? Auto-convert to text.
Sample Data Generator — FK-aware topological sort, 30 name heuristics (email → realistic email, price → 29.99), seeded reproducibility. Circular FK handling via deferred UPDATE.
Import from anywhere — MicroOLAP .pdd, DbSchema .dbs, Toad .dm2, plain SQL, live PostgreSQL via pg_catalog introspection.
CLI for CI/CD — pgdesigner lint, diff, generate, testdata, convert. Single binary, zero dependencies. Exit code 1 on lint errors — block merges with broken schemas.
Pricing
Free for non-commercial use — forever. All features, no time limit, no feature gates.
Commercial licenses start at $19 one-time for individuals. That’s less than a lunch with your team, and less than half what pgModeler charges.
What’s next
This is v1.0. Here’s what’s planned:
- Views and functions diff — currently tables, columns, indexes, FK, PK, constraints, enums, and partitions are diffed. Views and functions are next
- More sample data heuristics — expanding beyond 30 name patterns
- Demo mode — try PgDesigner in the browser without downloading
We’d love your feedback. Try it, break it, tell us what’s missing.