Policy

Cognition since 1.0 aow:Policy

Declarative constraints defining permissible, required, or restricted behavior.

States governance rules in a form that can be checked by software. Policies constrain behavior. They do not perform Actions.

Attributes

Attribute	Type	Required	Description
`rule_set`	text	yes	The rules, or a reference to them.
`rule_language`	string		The language rule_set is written in, for example "ODRL", "Rego", "DMN", "text".
`effect`	one of `permit`, `prohibit`, `require`, `escalate`		What the Policy does when it applies.
`trigger_conditions`	string-list		Conditions under which the Policy applies.
`severity`	one of `low`, `medium`, `high`, `critical`		Consequence of a violation.
`override_protocol`	text		Who may override the Policy, and how.

Every entity also has an id and a type, and may carry label, description, and the provenance facet: version, created_at, created_by, revision_of.

Relationships

Relationship	Points to	How many	Required	Meaning
`escalates_to`	Role, Human Actor	many		Where escalations go.

Pointed to by

From	Relationship	Read backward as
Intent, Plan, Task, Actor, Skill, Role	`constrained_by`	constrains
Feedback	`updates`	updated by
Guardian	`enforces`	enforced by

Notes

AOW does not prescribe a policy language. rule_set may hold rules in ODRL, Rego, Cedar, DMN, or plain text, named in rule_language.

Alignments

Standard or protocol	Term	Correspondence
W3C ODRL 2.2	`odrl:Policy`	closeMatch
OMG DMN	`Decision`	relatedMatch

Example

Source: examples/claims-processing.jsonld. The example passes all validation checks.

{
  "id": "policy-auto-approval",
  "type": "Policy",
  "label": "Auto-approval limit",
  "rule_set": "An eligibility decision on a disability claim under $50,000 may be recorded without human confirmation when eligibility confidence is at least 0.90.",
  "rule_language": "text",
  "effect": "permit",
  "trigger_conditions": [
    "claim_amount < 50000",
    "confidence >= 0.90"
  ],
  "severity": "high",
  "override_protocol": "Claims operations lead may lower the limit at any time; raising it requires model risk approval.",
  "escalates_to": [
    "role-claims-adjuster"
  ]
}

Term reference · JSON Schema · SHACL shape aowsh:PolicyShape