ai_guardrails

Creates, updates, deletes or gets an ai_guardrail resource or lists ai_guardrails in a region

Overview

Name	ai_guardrails
Type	Resource
Description	Definition of AWS::Wisdom::AIGuardrail Resource Type
Id	awscc.wisdom.ai_guardrails

Fields

get (all properties)

Name	Datatype	Description
assistant_id	string
assistant_arn	string
a_iguardrail_arn	string
a_iguardrail_id	string
name	string
blocked_input_messaging	string	Messaging for when violations are detected in text
blocked_outputs_messaging	string	Messaging for when violations are detected in text
description	string	Description of the guardrail or its version
▶topic_policy_config	object	Topic policy config for a guardrail.
▶content_policy_config	object	Content policy config for a guardrail.
▶word_policy_config	object	Word policy config for a guardrail.
▶sensitive_information_policy_config	object	Sensitive information policy config for a guardrail.
▶contextual_grounding_policy_config	object	Contextual grounding policy config for a guardrail.
tags	object
region	string	AWS region.

list (identifiers only)

Name	Datatype	Description
assistant_id	string
a_iguardrail_id	string
region	string	AWS region.

For more information, see AWS::Wisdom::AIGuardrail.

Methods

Name	Resource	Accessible by	Required Params
create_resource	ai_guardrails	INSERT	AssistantId, BlockedInputMessaging, BlockedOutputsMessaging, region
delete_resource	ai_guardrails	DELETE	Identifier, region
update_resource	ai_guardrails	UPDATE	Identifier, PatchDocument, region
list_resources	ai_guardrails_list_only	SELECT	region
get_resource	ai_guardrails	SELECT	Identifier, region

SELECT examples

get (all properties)

Gets all properties from an individual ai_guardrail.

SELECT
  region,
  assistant_id,
  assistant_arn,
  a_iguardrail_arn,
  a_iguardrail_id,
  name,
  blocked_input_messaging,
  blocked_outputs_messaging,
  description,
  topic_policy_config,
  content_policy_config,
  word_policy_config,
  sensitive_information_policy_config,
  contextual_grounding_policy_config,
  tags
FROM awscc.wisdom.ai_guardrails
WHERE
  region = 'us-east-1' AND
  Identifier = '{{ a_iguardrail_id }}|{{ assistant_id }}';

list (identifiers only)

Lists all ai_guardrails in a region.

SELECT
  region,
  a_iguardrail_id,
  assistant_id
FROM awscc.wisdom.ai_guardrails_list_only
WHERE
  region = 'us-east-1';

INSERT example

Use the following StackQL query and manifest file to create a new ai_guardrail resource, using stack-deploy.

Required Properties

/*+ create */
INSERT INTO awscc.wisdom.ai_guardrails (
  AssistantId,
  BlockedInputMessaging,
  BlockedOutputsMessaging,
  region
)
SELECT
  '{{ assistant_id }}',
  '{{ blocked_input_messaging }}',
  '{{ blocked_outputs_messaging }}',
  '{{ region }}';

All Properties

/*+ create */
INSERT INTO awscc.wisdom.ai_guardrails (
  AssistantId,
  Name,
  BlockedInputMessaging,
  BlockedOutputsMessaging,
  Description,
  TopicPolicyConfig,
  ContentPolicyConfig,
  WordPolicyConfig,
  SensitiveInformationPolicyConfig,
  ContextualGroundingPolicyConfig,
  Tags,
  region
)
SELECT
  '{{ assistant_id }}',
  '{{ name }}',
  '{{ blocked_input_messaging }}',
  '{{ blocked_outputs_messaging }}',
  '{{ description }}',
  '{{ topic_policy_config }}',
  '{{ content_policy_config }}',
  '{{ word_policy_config }}',
  '{{ sensitive_information_policy_config }}',
  '{{ contextual_grounding_policy_config }}',
  '{{ tags }}',
  '{{ region }}';

Manifest

version: 1
name: stack name
description: stack description
providers:
  - aws
globals:
  - name: region
    value: '{{ vars.AWS_REGION }}'
resources:
  - name: ai_guardrail
    props:
      - name: assistant_id
        value: '{{ assistant_id }}'
      - name: name
        value: '{{ name }}'
      - name: blocked_input_messaging
        value: '{{ blocked_input_messaging }}'
      - name: blocked_outputs_messaging
        value: '{{ blocked_outputs_messaging }}'
      - name: description
        value: '{{ description }}'
      - name: topic_policy_config
        value:
          topics_config:
            - name: '{{ name }}'
              definition: '{{ definition }}'
              examples:
                - '{{ examples[0] }}'
              type: '{{ type }}'
      - name: content_policy_config
        value:
          filters_config:
            - type: '{{ type }}'
              input_strength: '{{ input_strength }}'
              output_strength: null
      - name: word_policy_config
        value:
          words_config:
            - text: '{{ text }}'
          managed_word_lists_config:
            - type: '{{ type }}'
      - name: sensitive_information_policy_config
        value:
          pii_entities_config:
            - type: '{{ type }}'
              action: '{{ action }}'
          regexes_config:
            - name: '{{ name }}'
              description: '{{ description }}'
              pattern: '{{ pattern }}'
              action: null
      - name: contextual_grounding_policy_config
        value:
          filters_config:
            - type: '{{ type }}'
              threshold: null
      - name: tags
        value: {}

UPDATE example

Use the following StackQL query and manifest file to update a ai_guardrail resource, using stack-deploy.

/*+ update */
UPDATE awscc.wisdom.ai_guardrails
SET PatchDocument = string('{{ {
    "BlockedInputMessaging": blocked_input_messaging,
    "BlockedOutputsMessaging": blocked_outputs_messaging,
    "Description": description,
    "TopicPolicyConfig": topic_policy_config,
    "ContentPolicyConfig": content_policy_config,
    "WordPolicyConfig": word_policy_config,
    "SensitiveInformationPolicyConfig": sensitive_information_policy_config,
    "ContextualGroundingPolicyConfig": contextual_grounding_policy_config
} | generate_patch_document }}')
WHERE
  region = '{{ region }}' AND
  Identifier = '{{ a_iguardrail_id }}|{{ assistant_id }}';

DELETE example

/*+ delete */
DELETE FROM awscc.wisdom.ai_guardrails
WHERE
  Identifier = '{{ a_iguardrail_id }}|{{ assistant_id }}' AND
  region = 'us-east-1';

Permissions

To operate on the ai_guardrails resource, the following permissions are required:

Create

wisdom:CreateAIGuardrail,
wisdom:TagResource

Read

wisdom:GetAIGuardrail

Update

wisdom:UpdateAIGuardrail

Delete

wisdom:DeleteAIGuardrail

List

wisdom:ListAIGuardrails