File size: 10,848 Bytes
8ce0f27
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
# KYE™ Entity Types

This document lists the entity types KYE Protocol™ recognizes. Type names are stable identifiers used in entity records, policy inputs, and dictionaries.

This document defines **names only**. The lifecycle, validation rules, and attestation requirements for each type are part of the normative specification, which is published separately.

## Hierarchy

Every KYE™ entity resolves into one canonical containment tree (v3-final, locked 2026-05-14). The tree is parent-required at every node. Model, Tool, External App, and Audit Stream are **first-class Workspace-level entities** in v3 — they are not subtypes of Principal.

```
Tenant                                          (kye:tenant:…)    — billing + isolation root
  ├── Legal Entity                              (kye:lglent:…)    — registered company / regulator-facing legal person
  ├── Billing Account                           (kye:bill:…)      — Stripe customer + payment method + tax id
  ├── Domains                                   (kye:dom:…)       — verified DNS domains, branded URLs
  ├── Policies                                  (kye:pol:…)       — active Operating Model™, rule bundles, attestations
  └── Workspaces                                (kye:wsp:…)        — environment / dataspace (prod, sandbox, region-specific)
        ├── Projects                            (kye:prj:…)       — units of work; scoped budgets + quotas
        ├── Teams                               (kye:team:…)      — RBAC + ownership group
        ├── Resources                           (kye:res:…)       — D1 tables, R2 buckets, queues, connectors
        ├── Principals                          (kye:prin:…)      — the acting parties (Human / System / Agent)
        ├── Models                              (kye:model:…)     — inference artefacts; versioned
        ├── Tools                               (kye:tool:…)      — named callables; declared side-effects
        ├── External Apps                       (kye:extapp:…)    — 3rd-party integrations; partner connectors
        └── Audit Streams                       (kye:audstream:…) — append-only event destinations (SIEM, S3, Kafka)
```

Each entity carries a `state` field governed by a registered state machine (see `public/vocabulary/state-machines.md`). Every entity also carries a `state_machine_id` identifying which machine governs it. Relationships between entities are explicit typed edges (see `public/vocabulary/relationships.md`).

### Tenant-level siblings (5)

| Sibling | Stable type | ID class | Required parent | Purpose |
|---|---|---|---|---|
| Legal Entity     | `legal_entity`    | `lglent`    | `tenant_id` | registered legal person; regulator-facing record (company number, country, jurisdiction). One Tenant MAY hold multiple Legal Entities (group structure, foreign subsidiaries). |
| Billing Account  | `billing_account` | `bill`      | `tenant_id` | Stripe customer id, default payment method, billing email, tax id, currency. One Tenant MAY hold multiple Billing Accounts (split billing per business unit). |
| Domains          | `domain`          | `dom`       | `tenant_id` | verified DNS domains the Tenant owns (`acme.com`, `acme-uk.com`). Used for branded sender, federation routing, SSO. |
| Policies         | `policy_bundle`   | `pol`       | `tenant_id` | active Operating Model™, rule bundles, attestation policies, sub-processor list. One Tenant MAY hold multiple Policy Bundles (one per regime: PRA / DORA / EU AI Act). |
| Workspaces       | `workspace`       | `wsp`       | `tenant_id` | environment / dataspace. One Tenant MAY hold many Workspaces (prod / sandbox / eu-west-1 / consumer-bank / corporate-bank). |

### Workspace-level siblings (8)

| Sibling | Stable type | ID class | Required parent | Purpose |
|---|---|---|---|---|
| Projects      | `project`      | `prj`       | `workspace_id` | unit of work; scoped budget + quota; one Operating Model fragment per Project |
| Teams         | `team`         | `team`      | `workspace_id` | RBAC + ownership group; who-owns-what; bounded delegation surface; approval routing |
| Resources     | `resource`     | `res`       | `workspace_id` | D1 tables, R2 buckets, queues, connectors that live inside the workspace. Resources are leaves — they don't contain other entities |
| Principals    | `principal`    | `prin`      | `workspace_id`; `team_id` optional | the acting party — referenced as `actor_entity_id` / `principal_entity_id` in every Decision Map™ row |
| Models        | `model`        | `model`     | `workspace_id` | an inference artefact (LLM, vision model, embedding model); carries `provider`, `family`, `version`, `weights_hash`. First-class in v3. |
| Tools         | `tool`         | `tool`      | `workspace_id` | a named callable (function, MCP tool, API endpoint); carries `function_signature`, `declared_capabilities[]`, `side_effects[]`. First-class in v3. |
| External Apps | `external_app` | `extapp`    | `workspace_id` | a 3rd-party integration or partner connector acting on behalf of the tenant; carries `partner_org_id`, `connector_kind`, `oauth_subject`. First-class in v3. |
| Audit Streams | `audit_stream` | `audstream` | `workspace_id` | an append-only event-log destination the Workspace writes to (SIEM, S3, Splunk, Kafka topic); governed by an authority grant, not an IAM binding. New in v3. |

### Principal subtypes (v3: 4 subtypes)

In v3, `principal_class` has exactly four values — `human`, `system`,
`agent`, and `external_app` (the schema enum on
`kye.entity.principal.v1`, "reduced from six to four"). The v2 values
`model` and `tool` are **deprecated** — promoted to standalone entity
classes (`kye:model:…`, `kye:tool:…`). `external_app` is retained as a
principal_class (a 3rd-party integration acting on behalf of the tenant)
and ALSO exists as a standalone entity class (`kye:extapp:…`, see the
table above). See `constitution/DEVIATIONS.md` for the migration window.

Every Principal MUST resolve to exactly one subtype via its `principal_class` field:

| Subclass | Examples | Subclass-specific fields |
|---|---|---|
| `human`  | person, employee, contractor, operator, approver, auditor, regulator, customer, patient, support_user | full_name, email_hash, idp_subject, role |
| `system` | workload, service_account, api_client, serverless_function, container, vm, cluster, runtime, queue_consumer, webhook_endpoint, connector | service_principal_id, runs_in_workspace, owning_team_id |
| `agent`  | ai_agent, agent_runtime | model_entity_id (→ `kye:model:…`), tool_entity_ids[] (→ `kye:tool:…`), operating_loop_kind |
| `external_app` | 3rd-party integration / partner connector acting on behalf of the tenant | partner_org_id, connector_kind, oauth_subject |

Note that in v3 an `agent` Principal references its Model via a
`kye:model:…` entity ID and its Tools via `kye:tool:…` entity IDs —
it does not embed them as subtypes.

The hierarchy is enforced at schema time (`kye.entity.{tenant, legal_entity, billing_account, domain, policy_bundle, workspace, project, team, resource, principal, model, tool, external_app, audit_stream}.v1`) and at runtime by the PDP. KYE Onboarding Agent™ auto-creates a default Workspace + Team + admin Principal for every newly-granted Tenant so existing API keys, decisions, and audit rows always resolve to a valid parent chain.

## Human entities

| Type | Description |
|---|---|
| `person` | A natural person |
| `employee` | A person employed by a legal entity |
| `contractor` | A person engaged under a service contract |
| `operator` | A person operating a system or workflow |
| `approver` | A person empowered to approve actions |
| `auditor` | A person performing audit review |
| `regulator` | A regulatory authority representative |
| `customer` | A consumer of a product or service |
| `patient` | A patient in a healthcare context |
| `support_user` | A support-context user |

## Organizational entities

| Type | Description |
|---|---|
| `business` | A commercial organization |
| `legal_entity` | A registered legal entity |
| `subsidiary` | A subsidiary of a parent company |
| `parent_company` | A parent company |
| `department` | A department within an organization |
| `team` | A team within an organization |
| `public_authority` | A public-sector authority |
| `processor` | A data processor |
| `subprocessor` | A sub-processor |
| `supplier` | A supplier or vendor |
| `relying_party` | A relying party in an identity exchange |

## Technical entities

| Type | Description |
|---|---|
| `workload` | A running computational unit |
| `service_account` | A non-human service identity |
| `api_client` | An API client |
| `serverless_function` | A serverless function |
| `container` | A container instance |
| `vm` | A virtual machine |
| `cluster` | A compute cluster |
| `runtime` | A runtime environment |
| `connector` | An integration connector |
| `webhook_endpoint` | A webhook receiver |
| `queue_consumer` | A message-queue consumer |

## AI entities

| Type | Description |
|---|---|
| `ai_agent` | An AI agent |
| `agent_runtime` | An AI agent runtime |
| `model` | A model |
| `model_version` | A specific model version |
| `prompt_template` | A prompt template |
| `memory_store` | An agent memory store |
| `vector_index` | A vector index |
| `tool` | A tool or plugin |
| `guardrail` | A guardrail component |
| `evaluator` | An evaluator component |
| `workflow` | A multi-step workflow |

## Credential entities

| Type | Description |
|---|---|
| `key` | A cryptographic key |
| `api_key` | An API key |
| `jwt` | A JWT-formatted credential |
| `certificate` | An X.509 certificate |
| `x509_svid` | A SPIFFE X.509 SVID |
| `jwt_svid` | A SPIFFE JWT SVID |
| `verifiable_credential` | A W3C verifiable credential |
| `delegation_credential` | A delegation credential |
| `scope_credential` | A scope credential |
| `approval_credential` | An approval credential |

## Data and resource entities

| Type | Description |
|---|---|
| `dataset` | A dataset |
| `database` | A database |
| `table` | A database table |
| `file` | A file |
| `document` | A document |
| `record` | A record |
| `payment_instruction` | A payment instruction |
| `clinical_record` | A clinical record |
| `cardholder_object` | A cardholder data object (regulated) |
| `evidence_pack` | A compliance evidence pack |
| `policy_bundle` | A policy bundle |

## Naming rules

+ Type names are lowercase ASCII with underscores.
+ Type names are stable identifiers and must not be reused for unrelated semantics.
+ Adding a new type requires a vocabulary proposal. Mechanism details are out of scope for this document.