Selective Startup for Big Repos
How teams can work on one slice of a system without booting every service every time.
How teams can work on one slice of a system without booting every service every time.
A practical note from the TDK field guide for teams building service-heavy products without turning local development into infrastructure work.
Local should mean truthful
How teams can work on one slice of a system without booting every service every time. This article is part of the TDK field guide: practical notes for teams that want clearer specs, faster local feedback, and fewer development rituals.
The useful question is not whether local dev is fashionable. The useful question is whether it helps a developer make the next correct change with less waiting and less hidden context.
Speed only matters when people trust it
Specs, templates, and conventions are useful only when they stay attached to a working system. A clean document cannot prove that ports are assigned, credentials resolve, generated clients match services, or the local stack still boots.
That is why TDK keeps the practical layer close to the source of truth. The same service definition can inform local orchestration, generated files, generated diagrams, and assistant briefings.
Where TDK makes local work practical
TDK matters here because it turns service definitions into a runnable local surface. Docker, TypeScript, Tilt, Starlark, generated docs, credentials, and assistant context all become part of the same workbench instead of separate chores.
For selective startup for big repos, the point is simple: keep the system inspectable. A team should be able to understand what exists, run what matters, and carry the generated output forward without depending on opaque platform behavior.