AI Gateway · Schema

AIGatewayPolicy

A reusable AI gateway policy describing a guardrail, cost control, rate limit, or governance rule. Policies are attached to routes, tenants, or virtual keys to enforce content, cost, and access controls across LLM traffic.

AI GatewayLLM RouterLLM ProxyModel RoutingPrompt FirewallGuardrailsAI ObservabilityCost ControlsAI GovernanceAPI Gateway

Properties

Name Type Description
policyId string
name string
kind string Category of policy.
stage string Where this policy runs in the request lifecycle.
scope string What this policy applies to.
guardrail object Guardrail-specific configuration when kind=guardrail.
cost object Cost-control configuration when kind=cost.
rateLimit object Rate-limit configuration when kind=rate-limit.
access object Access-control configuration when kind=access.
dataResidency object Data-residency configuration when kind=data-residency.
audit object Audit-logging configuration when kind=audit.
enabled boolean
View JSON Schema on GitHub

JSON Schema

ai-gateway-policy-schema.json Raw ↑
{
  "$schema": "https://json-schema.org/draft/2020-12/schema",
  "$id": "https://raw.githubusercontent.com/api-evangelist/ai-gateway/refs/heads/main/json-schema/ai-gateway-policy-schema.json",
  "title": "AIGatewayPolicy",
  "description": "A reusable AI gateway policy describing a guardrail, cost control, rate limit, or governance rule. Policies are attached to routes, tenants, or virtual keys to enforce content, cost, and access controls across LLM traffic.",
  "type": "object",
  "properties": {
    "policyId": {
      "type": "string",
      "example": "pii-redaction-default"
    },
    "name": {
      "type": "string",
      "example": "PII Redaction"
    },
    "kind": {
      "type": "string",
      "description": "Category of policy.",
      "enum": ["guardrail", "cost", "rate-limit", "access", "data-residency", "audit", "transformation"],
      "example": "guardrail"
    },
    "stage": {
      "type": "string",
      "description": "Where this policy runs in the request lifecycle.",
      "enum": ["request", "response", "both", "background"],
      "example": "both"
    },
    "scope": {
      "type": "string",
      "description": "What this policy applies to.",
      "enum": ["global", "tenant", "team", "route", "virtual-key", "user"],
      "example": "tenant"
    },
    "guardrail": {
      "type": "object",
      "description": "Guardrail-specific configuration when kind=guardrail.",
      "properties": {
        "type": {
          "type": "string",
          "enum": ["pii", "prompt-injection", "jailbreak", "toxicity", "topic", "regex", "moderation", "secret-detection", "custom-webhook"],
          "example": "pii"
        },
        "categories": {
          "type": "array",
          "items": { "type": "string" },
          "example": ["ssn", "email", "phone", "credit-card"]
        },
        "action": {
          "type": "string",
          "enum": ["allow", "redact", "block", "warn", "log-only"],
          "example": "redact"
        },
        "engine": {
          "type": "string",
          "description": "Underlying engine implementing the guardrail.",
          "example": "presidio"
        }
      }
    },
    "cost": {
      "type": "object",
      "description": "Cost-control configuration when kind=cost.",
      "properties": {
        "limit": { "type": "number", "example": 1000.0 },
        "currency": { "type": "string", "example": "USD" },
        "period": {
          "type": "string",
          "enum": ["daily", "weekly", "monthly", "annual"],
          "example": "monthly"
        },
        "action": {
          "type": "string",
          "enum": ["alert", "throttle", "block"],
          "example": "block"
        }
      }
    },
    "rateLimit": {
      "type": "object",
      "description": "Rate-limit configuration when kind=rate-limit.",
      "properties": {
        "requestsPerMinute": { "type": "integer", "example": 600 },
        "tokensPerMinute": { "type": "integer", "example": 200000 },
        "concurrency": { "type": "integer", "example": 32 }
      }
    },
    "access": {
      "type": "object",
      "description": "Access-control configuration when kind=access.",
      "properties": {
        "roles": {
          "type": "array",
          "items": { "type": "string" },
          "example": ["platform-admin", "ai-developer"]
        },
        "allowedModels": {
          "type": "array",
          "items": { "type": "string" }
        },
        "deniedModels": {
          "type": "array",
          "items": { "type": "string" }
        }
      }
    },
    "dataResidency": {
      "type": "object",
      "description": "Data-residency configuration when kind=data-residency.",
      "properties": {
        "allowedRegions": {
          "type": "array",
          "items": { "type": "string" },
          "example": ["us-east-1", "us-west-2"]
        },
        "deniedProviders": {
          "type": "array",
          "items": { "type": "string" }
        }
      }
    },
    "audit": {
      "type": "object",
      "description": "Audit-logging configuration when kind=audit.",
      "properties": {
        "logRequestBody": { "type": "boolean", "example": true },
        "logResponseBody": { "type": "boolean", "example": true },
        "retentionDays": { "type": "integer", "example": 90 },
        "sink": { "type": "string", "example": "s3://ai-audit-logs/prod" }
      }
    },
    "enabled": {
      "type": "boolean",
      "example": true
    }
  },
  "required": ["policyId", "kind", "stage", "scope"]
}