Code generation
Skeema turns the parts of your design that are mechanical — database schemas and API scaffolds — into code that stays in sync with the diagram. Schema codegen is deterministic and free; API routes are derived from real flows.
Schema code (from ER boards)
Open the codegen drawer on any ER board to emit the schema in four targets. This is pure template code — no AI call, no cost, identical every time — so it’s safe to regenerate whenever the schema changes.
- •Prisma — models with camelCase fields,
@mapfor snake_case columns, enums, and inverse relations. - •PostgreSQL —
CREATE TABLEwithpgcryptoUUIDs, FK indexes, and ENUM types. - •MySQL — InnoDB engine, inline ENUMs,
utf8mb4. - •Mongoose — TypeScript interfaces plus schemas with
ObjectIdrefs.
API routes (from service nodes)
Derive API route scaffolds from a service node. Skeema picks the language and framework from the node’s declared tech stack and reads the project’s flows to shape the endpoints:
- •Language detected from tech — TypeScript, Python, Go, Java, Ruby, C#, Rust, or PHP.
- •Framework chosen to match — Express/Fastify/Nest, FastAPI, Gin, Spring, Rails, ASP.NET, Axum, Laravel.
- •Endpoints derived from real flow messages, not generic CRUD — and async-first when the service is connected to a stream.
Why deterministic matters
For schemas, determinism is a feature: the same diagram always produces the same code, so codegen is a reliable build step rather than a roll of the dice. AI is reserved for the genuinely ambiguous parts (route shaping from intent), where its judgment adds value.
- ✓Schema codegen (Prisma/PostgreSQL/MySQL/Mongoose) is deterministic, free, and always in sync.
- ✓API routes are language-aware and derived from real flows, not generic CRUD.
- ✓Round-trip: import schema → edit visually → generate clean code back out.