Authoring generators

Who this is for

You want generator behavior beyond what enrichments expose, or you're writing a new generator from scratch. You'll be reading and writing TypeScript in .skmtc/<project>/<gen-name>/src/.

None of this is required to use SKMTC — if you haven't used it as a consumer yet, start with using/. Authoring makes more sense once you've felt the customization limits, and it's more approachable than it sounds: generators build output with template literals (not ASTs) and compose with each other, so most authoring work is editing recognizable TypeScript. Stringable composition shows both ideas in real generator code.

The customization gradient

Use stock          → install + accept defaults
Configure          → enrichments in client.json
Customize behavior → clone + edit source     ← extending starts here
Author new         → write a new generator

Start here

• Anatomy of a generator — the one-page map of the parts (entry, base, enrichments, Projection, Snippets), how they run, and what keeps a generator maintainable. Read it before the tutorials.

Tutorials

Linear walkthroughs to learn by doing.

• 01 — Cloning a generator
• 02 — Authoring a model generator
• 03 — Authoring an operation generator

How-to guides

Recipes for specific problems.

• Change export paths
• Change identifier conventions
• Add a field type
• Swap a peer dependency
• Add enrichment options
• Compose with another generator
• Author a multi-generator stack
• Handle GraphQL instead of OAS

Recipes

Complete worked examples.

• Design system across many APIs
• Custom form field renderer

Shared resources you'll need

• Concepts — the DSL, the pipeline, cross-generator coordination, refs
• API reference — @skmtc/core exports, OAS object model, DSL classes
• CLI reference — clone, create, dev, bundle
• Stock generators reference — source layout and clone seams per generator
• Explanation — design rationale for the architecture you'll be extending