Deutschland-Stack knowledge graph

Unofficial prototype: the Deutschland-Stack as a knowledge graph.

The authoritative source is the official Tech-Stack Landkarte.

Building the knowledge graph

src/0-build-kg

Step	What it does
1. Fetch dataset	Fetches the compiled Landkarte dataset and its logos → data/upstream/ (full.json + metadata, logos.zip)
2. Reconstruct source	Reconstructs the landscape2 source files → data/reconstructed/ (landscape.yml + a minimal settings.yml)
3. Validate roundtrip	Structurally compares the rebuilt full.json against the authoritative one; to produce it, runs a full landscape2 build → data/scratch/build/

Running

npm install

npm run fetch # step 1
npm run reconstruct # step 2

# step 3
brew install cncf/landscape2/landscape2 # landscape2 CLI is required
npm run validate
landscape2 serve --landscape-dir data/scratch/build # optional: view the built site

Observations

A few things surfaced while working with the Landkarte's published artifacts:

• The two CSV files under Download (top right corner) are lossy exports of the build output (default landscape2 behaviour, not a BMDS addition), missing content compared to the landscape.yml and settings.yml they were built from. Those probably live in the repository the footer links to, which seems to have been public once but now returns a 404.
• The data model is inherited from the CNCF landscape and fits its new purpose imperfectly. Some fields are repurposed: license information appears under summary.release_rate and operating systems under summary.personas.
• Category and subcategory labels carry inconsistent leading whitespace — a mix of a normal space (U+0020) and an en-quad (U+2000) — so one category is split across several distinct strings. The reconstruction normalizes them to recover the four intended categories.

Disclaimer

• Not an authoritative source. This is an unofficial reconstruction of official artifacts.
• The Landkarte data (data/upstream/full.json) is content of the BMDS / Datenlabor BMI. No license is stated upstream; data/upstream/full.meta.json is kept as provenance documentation.
• The item logos are kept verbatim in data/upstream/logos.zip only to rebuild the site; no rights to them are claimed.
• Built with the help from AI coding tools; design decisions stay with the author, who reviews, understands and takes responsibility for every change.