Open Node specification
Which data a node exposes, which it does not, and how nodes talk to each other.
The specification says three things: which data a Node exposes, to whom, and in which form. It is the part of the system that must not be ours, because it is the part that decides whether a supply chain can talk to itself.
It is public for the same reason the manifesto says the artificial intelligence model is replaceable and the map is not: a company that builds its own node with other software stays interoperable, and no company ends up locked inside a vendor. Whoever adopts it strengthens the network, whoever they are.
What stays with Yempik is the engine: the computation, the governance of the rules and the log.
What a Node exposes, per view
Three views of the same company. Changing the view changes what you see, not where the number comes from.
| Data | Company | Buyer | Authority |
|---|---|---|---|
| Ninety-day capacity per family, with range | number | range | aggregated per supply chain |
| Committed load | order by order | percentage | aggregated |
| Supplier on-time performance | per supplier | overall | aggregated |
| Single-source dependencies | supplier, country, component | count and countries | count and countries, per supply chain |
| Non-EU share of value | per component | percentage | percentage, per supply chain |
| Open qualifications and engineering changes | detail | count and blocked families | aggregated |
| What we do not know | detail with decisions | "incomplete data on X" | aggregated |
| Log | complete | own requests | own requests |
| Prices, margins, customer names, drawings | yes | never | never |
Rules
- The Node belongs to the company: it shuts down when the company decides, and the knowledge stays with the company.
- No data leaves the Node except in the form the requested view allows, and every external request is logged.
- The language model is replaceable and receives only what the single operation needs.
- Every normalization or new rule is a proposal until a person with a name and a role approves it.
- Every answer cites the version of the data and of the rules it was produced with.
The tables are the public part. The engine that computes, governs and logs is Yempik's.
Where it will live
Being publishedThe specification will be published on GitHub next to cowork-os, under an open license, together with the Vertesa data generator. This page keeps the readable version, on the same numbering.
Open Node specification, version 0.1. Arsenale is a project by Yempik, Rome. The Node engine and the governance layer are Yempik's.