---
title: KYE Protocol™
emoji: ⛓️
colorFrom: green
colorTo: purple
sdk: static
pinned: true
license: apache-2.0
---

# KYE Protocol™ — Know Your Entity™

**Your model is on Hugging Face. Its authority isn't.**

The Hub shows the world what your model *can do* — weights, benchmarks,
model cards, lineage. It does not tell a bank, a hospital, or a
regulator whether they *may use it*. KYE Protocol™ answers that — and
proves it.

KYE Protocol™ is a runtime authority protocol for **delegated AI
actions**. It makes who/what acted — and on whose authority, with what
continuity of intent, under which scope, with what evidence —
observable, governable, revocable and replayable across humans, AI
agents, models, tools, workflows, APIs and autonomous services.

> Most systems prove **what happened**. KYE also asks whether the
> **meaning survived**.

---

## Capability vs. authority

| | Hugging Face — capability | KYE Protocol™ — authority |
|---|---|---|
| Question | What is this model and what can it do? | May you use it here, as this entity, for this purpose? |
| Artifacts | Weights · evals · model card · datasets | AI-BOM · obligation map · signed Evidence Pack™ |
| Consumer | Builders comparing models | Regulated buyers proving adoption |

You keep building on the Hub. KYE Protocol™ binds your model to the
obligations its use carries — so a regulated buyer can adopt it without
guessing. Read how model governance works end-to-end:
[kyeprotocol.com/huggingface.html](https://kyeprotocol.com/huggingface.html).

---

## Open contracts (Apache-2.0)

The protocol's contracts — vocabulary, taxonomies, schemas, examples —
are open source and free to adopt:

- **Source of truth:** [github.com/KYE-Protocol](https://github.com/KYE-Protocol)
- **Public landing:** [kyeprotocol.com](https://kyeprotocol.com)

| Hugging Face artifact | What | License |
|---|---|---|
| [`KYE-Protocol/schemas`](https://huggingface.co/datasets/KYE-Protocol/schemas) *(dataset)* | All KYE™ JSON Schemas (Core, Continuity, Discoverability, Ontology, Operating Model, Assurance Card, Formal Rules, Action Admissibility, Meaning Continuity) | Apache-2.0 |
| [`KYE-Protocol/reason-codes`](https://huggingface.co/datasets/KYE-Protocol/reason-codes) *(dataset)* | Canonical reason-code dictionary across decisions, drift events, obligations | Apache-2.0 |
| [`KYE-Protocol/examples`](https://huggingface.co/datasets/KYE-Protocol/examples) *(dataset)* | Validated example payloads for every schema | Apache-2.0 |
| [`KYE-Protocol/kye-models-governance`](https://huggingface.co/datasets/KYE-Protocol/kye-models-governance) *(article — also [on the HF blog](https://huggingface.co/blog/masoudmarandi/kye-models-governance))* | How KYE Protocol™ governs AI models — capability vs. authority, AI-BOM, declare → assess → certify | Apache-2.0 |
| [`KYE-Protocol/meaning-continuity-lab`](https://huggingface.co/spaces/KYE-Protocol/meaning-continuity-lab) *(Space)* | Interactive demo: change intent / interpretation / memory / handoff and watch the meaning-continuity score, drift reason-codes and admissibility recommendation update | Apache-2.0 |

## Governed datasets

Datasets are not just files — in KYE Protocol™ they are **governed data
assets**. Every dataset in this org is declared in the shape of the open
`kye.data_asset.v1` contract: classification `public`, licence
Apache-2.0, provenance pinned to the GitHub source of truth, and
CI-validated before every mirror push.

The same contracts govern *your* data at runtime: the
[KYE Data Governance Pack™](https://kyeprotocol.com/data-governance.html)
binds every agent data access (read, summarise, export, train_model,
share, disclose, …) to an active `kye.data_use_manifest.v1` —
default-deny on purpose, classification, residency and retention, with
one signed evidence row per access. The AI-BOM starts at the dataset:
governing the model without governing the data it was trained on is
half an answer.

## Paid runtime layer (proprietary)

The managed engines that operationalise the open contracts are
commercial:

- KYE Runtime Gateway™ · KYE Decision Engine™ · KYE Continuity Engine™ ·
  KYE Discovery Engine™ · KYE Ontology Engine™ · KYE Rights & Obligations
  Engine™ · KYE Admissibility Engine™ · KYE Meaning Continuity Engine™ ·
  KYE Meaning Drift Monitor™ · KYE Reconfirmation Flow™.

The procurement-pack briefing, sandbox, and onboarding live at
[kyeprotocol.com/engage.html](https://kyeprotocol.com/engage.html).

---

## Why "continuity of meaning"

In agentic systems, an action's authorised meaning can drift before
execution. KYE Protocol™ does not assume meaning survives delegation
automatically. It checks whether **original intent, constraints,
context, memory, incentives, timing and state** remain aligned across
handoffs before the action is admitted, authorised, committed and
evidenced.

That check is `KYE Meaning Continuity™` — try it interactively in the
[KYE Meaning Continuity Lab™ Space](https://huggingface.co/spaces/KYE-Protocol/meaning-continuity-lab).

---

KYE™, KYE Protocol™, Know Your Entity™, and all `KYE *` profile and
engine names are trademarks of the KYE Protocol™ project.