One runtime.
Every business
application.
A metadata-first runtime that compiles entity definitions into real Postgres schemas, REST + GraphQL APIs, generated UI, and row- and column-level security. Multi-tenant by default. Cloud or self-hosted. AI-native by design.
Compile metadata.
Get a real app.
No codegen, no scaffolding, no boilerplate to babysit. Metadata goes in. A complete, production-grade application comes out — every layer, observable end to end.
› dforge compile metadata/*.dmod✓ parse 14 entities · 38 views · 9 roles✓ validate 0 errors · 0 conflicts✓ schema postgres migrations applied✓ api REST + GraphQL + JSON-RPC✓ ui list · form · kanban · calendar✓ rls row & column policies installed✓ audit change history enabled✓ i18n 3 locales registered› ready · uptime 100% · 12 ms p95
One database
per tenant.
No exceptions.
Schema-per-tenant isolation, not row-level "tenant_id" filtering. Each organization gets its own Postgres schema, its own backups, its own migrations. Real isolation. Compliance-friendly.
Composable security.
Declared, not coded.
Every access rule is metadata. Versioned, reviewable, auditable. No more security logic scattered across controllers and stored procedures.
Composable, additive roles. A user with two roles inherits the union of their rights.
role Sales can: read, update.own
role Manager extends Sales
can: delete, export Filter rows declaratively per role. Policies live with the metadata, not in app code.
policy OwnRecordsOnly on Lead
for Sales
where owner = $user Hide or mask fields per role without forking the schema or building view layers.
field Lead.salary
visible: HR, Finance
masked: Sales Organizational boundaries map to data scopes. Region heads see only their region.
scope Region.EU
grants: read, write
to: Manager.eu Describe it.
AI builds
the module.
Run npx @dforge/create-module, open in Claude Code, and describe what you need. The AI skill knows every field type, every convention, every DSL function. Or import your existing SQL / DBML schema and let AI finish the rough cut. Validate locally, package, install.
learn how modules work → Import your schema.
Get a working business app.
dForge meets developers where they already work. Point the importer at your existing database, open the module in your editor, and ship — with first-class tooling, real APIs, and metadata you can read, diff, and review.
A VS Code extension with DSL syntax highlighting and JSON-Schema validation across every module file. Zed highlighting today — more editors on the way.
Feed a SQL / DBML schema to the importer and get a ready-to-install module — entities, views, menus, folders and roles inferred for you. Your legacy data comes along.
Scaffold, validate, package and install from the terminal — or drive the whole module surface from your AI agent through the dForge MCP server.
Run it our way,
or yours.
Spin up a workspace in seconds. We handle infra, backups, scaling, and updates. EU and US regions available.
› signup → workspace.dforge.app
› install crm hr fin
✓ ready in 47s Run dForge inside your infra. Docker, Kubernetes, or bare metal. Bring your own Postgres. Your data never leaves your network.
$ docker compose up dforge
$ dforge tenant create acme
✓ ready · acme.local Built for
every language.
Field labels, menu titles, error messages, and validation strings — all live as metadata. Add a locale, ship instantly. Tenants pick their own languages independently.
Less plumbing.
More platform.
Open a workspace on dforge.app or talk to us about a hands-on walkthrough with your team.