Policy: AWS > Bedrock > Guardrail > Settings > Content Policy > Prompt Attack Filter Input Strength

Filter strength applied to prompt-injection attacks in model input. Defaults to the AWS-recommended baseline (HIGH); set to NONE to disable. The PROMPT_ATTACK filter is included in the enforced filtersConfig when this is non-NONE; if NONE, the filter is removed from the guardrail entirely.

AWS forces outputStrength to NONE for the PROMPT_ATTACK filter, so there is no matching output strength sub-policy.

Targets

This policy targets the following resource types:

AWS > Bedrock > Guardrail

Primary Policy

This policy is used with the following primary policy:

AWS > Bedrock > Guardrail > Settings > Content Policy

Controls

Setting this policy configures this control:

AWS > Bedrock > Guardrail > Settings

Policy Packs

This policy setting is used by the following policy packs:

Enforce Content Filtering for AWS Bedrock Guardrails

Policy Specification

Schema Type	`string`
Default	`HIGH`
Valid Values [YAML]	`NONE` `LOW` `MEDIUM` `HIGH`
Examples [YAML]	HIGH

In Your Workspace

Policy Settings by Type report

Developers

Category URI

tmod:@turbot/turbot#/control/categories/security

Policy Type URI

tmod:@turbot/aws-bedrock#/policy/types/bedrockGuardrailSettingsContentPolicyPromptAttackFilterInputStrength

GraphQL

query policyType(id: "tmod:@turbot/aws-bedrock#/policy/types/bedrockGuardrailSettingsContentPolicyPromptAttackFilterInputStrength") { … }

query policySettings(filter: "policyTypeId:'tmod:@turbot/aws-bedrock#/policy/types/bedrockGuardrailSettingsContentPolicyPromptAttackFilterInputStrength'") { … }

query policyValues(filter: "policyTypeId:'tmod:@turbot/aws-bedrock#/policy/types/bedrockGuardrailSettingsContentPolicyPromptAttackFilterInputStrength'") { … }

CLI

Get Policy Type

turbot graphql policy-type --id "tmod:@turbot/aws-bedrock#/policy/types/bedrockGuardrailSettingsContentPolicyPromptAttackFilterInputStrength"

Get Policy Settings

turbot graphql policy-settings --filter "policyTypeId:tmod:@turbot/aws-bedrock#/policy/types/bedrockGuardrailSettingsContentPolicyPromptAttackFilterInputStrength"