April 12, 2026 (modified: 2026-05-23)

Internal Developer Platform (IDP) Part 1 - A Walking Skeleton

Motivation

Platform engineering book examples rot fast: images get renamed, Helm charts break, a year later it doesn’t run anymore and requires substantial effort to make it work again. This series builds a walking skeleton IDP from scratch that stays verifiable over time — fully automated setup, no cloud account required, and runs on a single laptop.

The platform targets software engineers who want a concrete, working IDP reference they can clone, study, and carry toward production. AI assisted the implementation throughout; this series documents both the platform and what that collaboration looked like in practice.

The source is at codeberg.org/michugfeuer/idp-on-k8s.

Objectives

  • Serves as IDP reference and blueprint, illustrating core IDP functionality

  • Provides a fully functional walking skeleton covering all relevant aspects of an IDP

  • Does not require to be exposed publicly

  • Reduces bit rot through complete automated setup and testing

  • Uses cloud native components

  • Follows cloud native recommended practices

  • Runs on a single machine with modest spec, developed on a Linux-notebook from 2014 with 8 GB RAM and 4 CPU

  • Based on Kubernetes, independent of specific implementation as much as possible, documented if vendor specific features have to be used.

  • Self contained on Kubernetes

  • Targets production readiness; documents gaps where resource constraints apply

  • Based on and exemplifies the IDP reference architecture with the 5 planes: Developer Control Plane, Integration & Delivery Plane, Resource Plane, Observability Plane, and Security Plane.

  • Suits workshops, presentations, and self-study

  • Stays as simple as needed, no simpler

  • Keeps all components loosely coupled and replaceable

  • Links implementation decisions back to platform engineering, DevOps, DevSecOps

Prerequisites

The project and the accompanying blog posts assume that the reader is familiar with Kubernetes and cloud native principles.

This post and the underlying code was written and developed in collaboration with several AIs.

Tags: AI-assisted Claude IDP Platform Engineering