Policy Engine¶

The Policy Engine evaluates request context against YAML-defined policy rules. It supports tenant-level, use-case-level, role-based, and channel-specific policies.

Policy Structure¶

name: policy_name
version: "1.0.0"
description: What this policy set does
tenant_id: optional-tenant-id
rules:
  - id: rule-id
    description: What this rule checks
    scope: global | tenant | use_case | channel | role
    condition:
      field: value
      field2: { "$gt": 100 }
    decision: allow | deny | warn | escalate
    priority: 10  # lower = higher priority

Condition Operators¶

Operator	Example	Meaning
equality	`field: value`	Exact match
`$in`	`field: { "$in": ["a", "b"] }`	Value in list
`$gt`	`field: { "$gt": 100 }`	Greater than
`$lt`	`field: { "$lt": 50 }`	Less than
`$ne`	`field: { "$ne": "blocked" }`	Not equal

Example Policies¶

See policies/examples/ for healthcare, finance, and general tenant policies.

Endpoint¶

POST /v1/policies/evaluate

Source¶

src/agentguard/policy/