What this repository is
Nokto is an open protocol for receiving accounts. This document states its scope: what belongs to the protocol, and what is built above it.
Scope
This repository is the protocol: the on-chain mechanism and the interfaces to it. Compliance orchestration, payment rails, underwriting, and operator consoles are not part of the protocol — they are what you build on top of it. Nokto Platform builds one such product. The protocol does not depend on it, and never will.
Something belongs here when it executes on-chain, when a counterparty needs it to verify a claim, or when a third-party integrator or plugin author needs it to build.
The guarantee behind the last sentence
Nokto Platform maintains this protocol and builds a commercial product on it. That
product resolves @nokto-protocol/schema, @nokto-protocol/sdk, and
@nokto-protocol/plugin-kit from the public npm registry at pinned versions —
the same packages, the same versions, resolved the same way you would resolve
them. Never by relative path, git dependency, or submodule.
These packages are not published yet, so today this is a commitment rather than something you can check. It is enforced on the maintainer's side by CI that rejects local-path resolution, and it becomes checkable by anyone the moment the first version publishes.
That is a rule about the maintainer, not about you, and it is the only thing that makes the scope statement checkable rather than merely stated. If the maintainer's product could reach past the published interface, the published interface would not be the real one, and everyone building against it would be building against a fiction. The rule exists so that does the protocol work without the product? has a factual answer.
Four boundaries worth working through
Each splits in the same place: the mechanism is the protocol, the judgement is not.
| The protocol | Built above it | |
|---|---|---|
| Assurance tiers | What tier 1 means. A verifier cannot interpret a proof without it, and a tier it cannot interpret is worthless. | How a would-be issuer is assessed, what evidence it must produce, and who signs that off. |
| Issuer registry | The IssuerRecord layout, the key-id derivation, and the tier bounds. These are consensus rules; anyone auditing the chain needs them. | The decision to admit a particular issuer at a particular tier. |
| Predicates | discloseIncomeAtLeast and the shape of future predicates. A verifier pins one and must keep working. | What a lender concludes from the answer — the model, the cut-offs, the pricing. That belongs to the lender. |
| Entitlement | The conservation rule: a source cannot promise away more than it earns. A circuit constraint. | Which jurisdiction's rules a corporate structure must satisfy, and how a weight is derived from a shareholders' agreement. |
Reading the right-hand column as a list of things withheld gets it backwards. It is a list of layers this protocol deliberately does not occupy, so that more than one party can occupy them.
The failure this is written against
The tempting mistake is not omitting a secret. It is publishing a mechanism that only makes sense alongside a policy you cannot see — so you can read the code, cannot reproduce the behaviour, and conclude the open packages are decorative. If a published interface cannot be used to build something real without asking us for the missing half, it is the wrong shape, not the wrong amount of documentation. Say so; that is a bug report we will act on.
Permanence
Everything committed here is public, permanently. A force-push does not un-publish what was cloned, forked, or indexed — and that applies to commit messages and issue templates as much as to code. Before opening a pull request, see CONTRIBUTING.md.