AWS nova medium security documentation change
Summary
Removed detailed response structure documentation including security-related 'content_filtered' stop reason and other response metadata
Security assessment
The removal of the 'content_filtered' stop reason documentation obscures awareness of automated content moderation, a security control that prevents policy-violating outputs. This reduces transparency about security safeguards in the API response handling.
Diff
diff --git a/nova/latest/nova2-userguide/using-invoke-api.md b/nova/latest/nova2-userguide/using-invoke-api.md index 6c9b8866b..2286b74e0 100644 --- a//nova/latest/nova2-userguide/using-invoke-api.md +++ b//nova/latest/nova2-userguide/using-invoke-api.md @@ -5 +5 @@ -Request structureRequest parametersResponse structure +Request structureRequest parameters @@ -9,12 +9 @@ Request structureRequest parametersResponse structure -The Invoke API provides direct access to Amazon Nova models with full control over the request and response format. Unlike the Converse API which abstracts model-specific details, the Invoke API allows you to work directly with the model's native request and response structures. - -The Invoke API is useful when you need: - - * Direct control over request formatting - - * Access to model-specific response fields - - * Custom handling of binary data encoding - - - +The Invoke API provides direct access to Amazon Nova models with more ability to control the request and response format. Unlike the Converse API which abstracts model-specific details, the Invoke API allows you to work directly with the model's native request and response structures. @@ -76,45 +64,0 @@ The Invoke API supports the following key parameters: -## Response structure - -The response includes the model's output and metadata: - - - { - "output": { - "message": { - "role": "assistant", - "content": [ - { - "text": "Machine learning is a subset of artificial intelligence..." - } - ] - } - }, - "stopReason": "end_turn", - "usage": { - "inputTokens": 15, - "outputTokens": 128, - "totalTokens": 143 - } - } - -Possible `stopReason` values: - - * `end_turn`: Logical end of the turn was reached - - * `max_tokens`: Output reached the maxTokens limit - - * `content_filtered`: Output violated Amazon Responsible AI policy - - * `malformed_model_output`: Model produced invalid output - - * `malformed_tool_use`: Model produced invalid tool use output - - * `service_unavailable`: Built-in tool service could not be reached - - * `invalid_query`: Query sent to built-in tool was invalid - - * `max_tool_invocations`: Built-in tool did not produce valid result after retries - - - -