AWS Bedrock (AssumeRole)Install

Summary

Send a prompt to a specified AWS Bedrock model using AssumeRole authentication, optionally including an image alongside text, and return the model's response.

Inputs

• Model ID (required) — Select the AWS Bedrock model ID to use for the request. When Application Inference Profiles JSON is configured in the credentials, only models present in that JSON are listed here.
• Input Type (required) — Choose whether to send only text or a combination of text and image to the model.
• Image Binary Property — Specify the name of the binary property containing the image to analyze; required if Input Type is 'Text and Image'.
• Prompt (required) — The text prompt to send to the model.
• Max Tokens (required) — Maximum number of tokens to generate in the model's response.
• Temperature — Controls the randomness of the response, ranging from 0.0 (more deterministic) to 1.0 (more random).

Output shape

a list of JSON objects, one per input item, each containing the Bedrock model ID used, the prompt sent, the raw API response, usage metadata, the text content of the model's completion, and a timestamp.

Supports batch processing of multiple input items. When Input Type is 'Text and Image', the node requires the referenced binary image data on the input item. Uses temporary credentials acquired via AWS STS AssumeRole from configured base credentials. Only models with 'anthropic.claude' prefix are supported. Throws errors on missing required inputs or unsupported models unless 'Continue On Fail' is enabled.