Every config your stack needs. Already written.

35+ product capabilities, all driven by one service.json — which also generates 12-17 files per service and 40+ stack-wide artifacts. Powered by Tilt and Starlark, running on plain Docker. Zero Kubernetes knowledge required.

The two weeks nobody budgets for

This is not a developer problem. It is a payroll problem. Here is how the first month of a new project actually goes — and how it goes with TDK.

Day one — 5 offers signed at once

Hiring one developer per week is the smart advice. In reality, projects hire the whole starting team together — 5 developers, one Monday, full payroll running.

5 salaries · day 1

Day two — "how do I run this?"

Nobody can start. The business analyst hasn't finished the specs. The .env files live in someone's inbox. The setup wiki was last edited eight months ago and describes a stack that no longer exists.

5 salaries · waiting

Week one — the environment lottery

One developer gets the backend running but not the database. Another has the database but not the events queue. Two are still installing tooling. None of them can see the whole system.

5 salaries · still waiting

Week two — first commit, maybe

By the time everyone can run the project, two weeks of 5 salaries are gone — burned on waiting, not on building. This is the most expensive line item nobody put in the budget.

5 salaries × two weeks · burned

With TDK — day one looks different

Clone the repo. Run tdk up. Every new hire has the entire system — databases, queues, frontends — running on their own laptop in minutes. Before the first BA call even ends, all five are writing code.

minutes · not weeks

Configuration generation

One service.json becomes 12-17 production-grade files per service - written correctly and kept in sync.

01

TypeScript config

tsconfig.json with proper paths, includes and compiler options for backend, frontend and libraries.

02

Vite configuration

vite.config.ts for local, Docker and production builds with optimized settings.

03

Docker and Compose

Multi-stage Dockerfile with golden image layering. Compose orchestration per service.

04

nginx proxy

Reverse proxy rules with Traefik integration for pretty local URLs.

05

Environment files

.env files with service-specific variables, database URLs and internal dependencies.

06

Package manager

.npmrc and bunfig.toml wired to the local Verdaccio registry and npm.

07

AGENTS.md guard

Generated AGENTS.md keeps AI coding assistants from editing autogenerated files.

Development experience

Built on Tilt — 9.4k GitHub stars — for a feedback loop measured in seconds.

08

Hot reload

Instant code sync with Tilt live_update. No container rebuilds while you type.

09

Selective startup

Run 5 services instead of 100. Start only what your task touches.

10

Auto-discovery

Finds every service.json in your directory tree. Zero manual registration.

11

Manifest synthesis

Services without manifests auto-configure from directory structure.

12

Config-gen resources

Every manifest gets a visible Tilt resource with its own status, logs and errors.

13

Dependency graph

internalDependencies resolved automatically. Services start in the correct order.

14

Stack grouping

Boot entire feature sets with one command: tdk up --stack api.

Infrastructure

Production-grade Docker architecture without a production-grade ops team.

15

Golden images

Four cached layers — L1 OS, L2 runtime, L3 builder, L4 app — shared across all services.

16

Docker only

Pure Docker + Tilt. No cluster, no control plane, no YAML maze.

17

Traefik integration

Pretty URLs like user.backend.localhost, configured automatically.

18

API gateway

Routing between services with path-based rules, generated from manifests.

19

Local registry

Verdaccio npm registry for sharing libraries between services.

20

Playwright testing

E2E configuration per service, ready to run against the live stack.

21

C4 diagrams

Architecture diagrams generated from your actual service topology.

Database and state

Per-service PostgreSQL, migrations, cache and events — provisioned, not documented.

22

Database provisioning

Isolated PostgreSQL database per service with proper naming.

23

Prisma integration

Prisma config with DATABASE_URL generated. Migrations built in.

24

Readiness checks

Services wait for the database. No connection errors on boot.

25

Migration scripts

prisma migrate runs on service startup, automatically.

26

Persistent volumes

Data survives container restarts during development.

27

NATS JetStream

Event-driven pub/sub wired between services out of the box.

28

Redis support

Shared Redis for caching and sessions, auto-configured.

Security and operations

The boring, expensive parts — validated before they can break your morning.

29

Infisical secrets

Machine identities, paths and dynamic secrets per service.

30

Pre-flight validators

Manifests validated before build. Misconfigurations caught early, in plain language.

31

Dependency manifest

A full service graph for complex trees, resolved automatically.

32

Path safety

Docker-safe path handling. No more works-on-my-machine builds.

33

Health checks

Tilt monitors every service and restarts on failure.

34

Library autodiscovery

Local @tdk/* packages auto-linked from package.json.

35

Resource aliases

Reference services by name in dependencies and commands.

Day one should ship code

Join the waitlist and give your next five hires a running system in minutes — not two weeks of waiting.

Join waitlist