AWS bedrock documentation change
Summary
Updated terminology and formatting for guardrail response examples, changed action type from MASKED to ANONYMIZED, and clarified output scope behavior
Security assessment
Changes improve documentation of security controls (guardrails) but do not indicate a specific security vulnerability being addressed. The ANONYMIZED action change reflects terminology refinement rather than a security fix.
Diff
diff --git a/bedrock/latest/userguide/guardrails-use-independent-api.md b/bedrock/latest/userguide/guardrails-use-independent-api.md index 1a87622dc..2f45fa3e7 100644 --- a//bedrock/latest/userguide/guardrails-use-independent-api.md +++ b//bedrock/latest/userguide/guardrails-use-independent-api.md @@ -170 +170 @@ The outputs of the `ApplyGuardrail` request depends on the action guardrail took -No guardrail intervention +Guardrails takes no action @@ -173 +173 @@ No guardrail intervention -Request example +**Request example** @@ -180 +180 @@ Request example - "text": "Hi, my name is Zaid. Which car brand is reliable?", + "text": "Hi, my name is Zaid. Which car brand is reliable?" @@ -185 +185 @@ Request example -Response if guardrails did not intervene +**Response example** @@ -200 +200 @@ Response if guardrails did not intervene -Guardrails intervened with BLOCKED action +Guardrails blocks content @@ -203 +203 @@ Guardrails intervened with BLOCKED action -Response example +**Response example** @@ -215 +215 @@ Response example - "text": "Configured guardrail canned message, i.e cannot respond" + "text": "Configured guardrail canned message (i.e., can't respond)" @@ -236 +236 @@ Response example -Guardrails intervened with MASKED action +Guardrails masks content @@ -239 +239 @@ Guardrails intervened with MASKED action -Response example +**Response example** @@ -241 +241 @@ Response example -Guardrails intervened with name masking (name is masked) +Guardrails intervenes by masking the name `ZAID`. @@ -264 +264 @@ Guardrails intervened with name masking (name is masked) - "action": "MASKED" + "action": "ANONYMIZED" @@ -271 +271 @@ Guardrails intervened with name masking (name is masked) -CLI Example +AWS CLI example @@ -274 +274 @@ CLI Example -Input example +**Input example** @@ -277 +276,0 @@ Input example - # Make sure preview CLI is downloaded and setup @@ -294,2 +293 @@ Input example - -Output example +**Output example (blocks content)** @@ -331 +329 @@ Content is considered detected if it breaches your guardrail configurations. For -By default, the [ApplyGuardrail](https://docs.aws.amazon.com/bedrock/latest/APIReference/API_runtime_ApplyGuardrail.html) operation only returns detected content in a response. You can specify the `outputScope` field with the `FULL` value to return the full output. The response will also include non-detected entries for enhanced debugging. +By default, the [ApplyGuardrail](https://docs.aws.amazon.com/bedrock/latest/APIReference/API_runtime_ApplyGuardrail.html) operation only returns detected content in a response. You can specify the `outputScope` field with the `FULL` value to return the full output. In this case, the response will also include non-detected entries for enhanced debugging.