TODL

@pragmatic-tech-ai/todl — the typed-object language: a language, meta-model system, and model compiler over a reflective typed graph (load → validate → emit). ESM, strict TypeScript.

• Architecture — how the whole system fits together, in fifteen sections. Each section links to a full deep-dive article (see below).
• The TODL language — the language reference: types, meta-models, taxonomies, operators, and the model compiler.

Built on todl-runtime.

Architecture deep dives

The architecture overview is the map; these are the detailed articles behind each section, for engineers who need the full picture of a subsystem.

• What TODL is — the problem, the workflow, the three defining properties.
• The two-minute mental model — the pipeline diagram, walked slowly.
• Repository map — a guided tour of the source tree.
• Core concepts — the typed graph, declarations, namespaces, classes, bases.
• The compiler — parse, load, validate, emit, and the public check API.
• Manifest and reflection — the binary format and the runtime read API.
• Consuming a model — model-data, reflection-client, codegen, authoring.
• Publish and packages — the compile-and-persist spine and resolution chain.
• Projects and solutions — project types, closure reassembly, solution builds.
• The build system — the generic engine and the two build flavours.
• Project content generators — how a project’s editable src/ app and generated/ DTO are produced, and the build’s require-never-create boundary.
• The runnable app — from index.html to a painted UI.
• Tooling — the language server, analysis service, CLI, and migrator.
• Package surface and dependencies — the export strategy and the dependency graph.
• Testing and conventions — the runner, smoke tests, goldens, house style.
• End-to-end walkthroughs — four complete journeys through the pipeline.