The rivano.policies resource lets you manage declarative governance rules programmatically. Policies fire on every proxied request or response and take one of three actions: block, redact, or warn.

List policies

import Rivano from '@rivano/sdk';

const rivano = new Rivano({ apiKey: 'rv_...' });

const { data, total } = await rivano.policies.list();
for (const policy of data) {
  console.log(`${policy.name} [${policy.onPhase}] → ${policy.action} (${policy.enabled ? 'on' : 'off'})`);
}

import requests

response = requests.get(
    'https://api.rivano.ai/api/policies',
    headers={'Authorization': 'Bearer rv_...'},
)
for policy in response.json()['data']:
    print(policy['name'], policy['action'])

curl https://api.rivano.ai/api/policies \
  -H "Authorization: Bearer rv_..."

Create a policy

import Rivano from '@rivano/sdk';

const rivano = new Rivano({ apiKey: 'rv_...' });

const policy = await rivano.policies.create({
  name: 'block-prompt-injection',
  description: 'Block requests with injection score >= 0.7',
  onPhase: 'request',
  conditionType: 'injection_score',
  conditionConfig: { threshold: 0.7 },
  action: 'block',
  enabled: true,
});

console.log('Policy created:', policy.id);

import requests

response = requests.post(
    'https://api.rivano.ai/api/policies',
    headers={
        'Authorization': 'Bearer rv_...',
        'Content-Type': 'application/json',
    },
    json={
        'name': 'block-prompt-injection',
        'onPhase': 'request',
        'conditionType': 'injection_score',
        'conditionConfig': {'threshold': 0.7},
        'action': 'block',
        'enabled': True,
    },
)
print(response.json()['id'])

curl https://api.rivano.ai/api/policies \
  -X POST \
  -H "Authorization: Bearer rv_..." \
  -H "Content-Type: application/json" \
  -d '{
    "name": "block-prompt-injection",
    "onPhase": "request",
    "conditionType": "injection_score",
    "conditionConfig": {"threshold": 0.7},
    "action": "block",
    "enabled": true
  }'

Create parameters

Parameter	Type	Required	Description
`name`	`string`	Yes	Unique policy name
`description`	`string`	No	Human-readable description
`onPhase`	`'request' \| 'response'`	Yes	When to evaluate: before or after the LLM call
`conditionType`	`string`	Yes	See condition types below
`conditionConfig`	`object`	Yes	Configuration for the condition
`action`	`'block' \| 'redact' \| 'warn'`	Yes	What to do when the condition is met
`enabled`	`boolean`	No	Whether the policy is active (default: `true`)
`teamId`	`string`	No	Scope policy to a specific team

Condition types

conditionType	Phase	conditionConfig fields	Description
`injection_score`	`request`	`threshold: number` (0–1)	Block/warn if injection risk score ≥ threshold
`pii_detected`	`request` or `response`	`types?: string[]`	Fire if PII is detected (email, phone, SSN, etc.)
`token_count`	`request`	`maxTokens: number`	Fire if estimated token count exceeds limit
`model_name`	`request`	`allowlist?: string[]`, `denylist?: string[]`	Fire if model is not in allowlist or is in denylist

Update a policy

import Rivano from '@rivano/sdk';

const rivano = new Rivano({ apiKey: 'rv_...' });

const updated = await rivano.policies.update('policy_abc123', {
  enabled: false,
  conditionConfig: { threshold: 0.6 },
});

Delete a policy

await rivano.policies.delete('policy_abc123');

Policy templates

Rivano ships a set of curated policy templates. List them or apply a full template pack:

List available templates

import Rivano from '@rivano/sdk';

const rivano = new Rivano({ apiKey: 'rv_...' });

const templates = await rivano.policies.templates();
for (const t of templates) {
  console.log(t.pack, t.name, t.action);
}

Apply the foundational pack

import Rivano from '@rivano/sdk';

const rivano = new Rivano({ apiKey: 'rv_...' });

// Applies 4 policies at once:
// - block-injection (request, block)
// - redact-pii-response (response, redact)
// - warn-large-prompt (request, warn, >4000 tokens)
// - block-ssn (request, block)
const policies = await rivano.policies.applyTemplate('foundational');
console.log(`Applied ${policies.length} policies`);

import requests

response = requests.post(
    'https://api.rivano.ai/api/policies/templates/apply',
    headers={
        'Authorization': 'Bearer rv_...',
        'Content-Type': 'application/json',
    },
    json={'pack': 'foundational'},
)
policies = response.json()
print(f"Applied {len(policies)} policies")

curl https://api.rivano.ai/api/policies/templates/apply \
  -X POST \
  -H "Authorization: Bearer rv_..." \
  -H "Content-Type: application/json" \
  -d '{"pack": "foundational"}'

💡

The foundational pack is the fastest way to get meaningful governance in place. You can customize any of the applied policies afterward via policies.update().

Error handling

Error	When it occurs
`SdkAuthError`	Invalid API key
`SdkNotFoundError`	Policy ID does not exist
`SdkError`	Invalid condition type or action combination

CLI Policies — Apply templates and manage policies from the terminal
SDK Governance — Posture scores and compliance tracking
Core Concepts — Policies — How policies work
Gateway Middleware — How policies are evaluated in the pipeline

SDK — Policies