aws.bedrock.chat

Send a conversation to a Bedrock model via the Converse API

Send a conversation to an Amazon Bedrock model and receive the assistant reply.

Uses the Bedrock Converse API, which works with all Bedrock-supported models (Anthropic Claude, Amazon Titan, Meta Llama, Mistral, etc.).

Inputs:

messages: List of {role, content} objects (required). content can be a string or a Bedrock-native content list.
model_id: Bedrock model ID (optional, default: anthropic.claude-3-haiku-20240307-v1:0).
system: System prompt string (optional).
max_tokens: Maximum tokens in the response (optional, default: 1024).

Outputs (terminal state_data):

Plugin required: context.get_plugin("aws") must expose .bedrock_client().

Overview

Field	Type	Required	Default	Description
`messages`	json	Yes	—	List of {role, content} message objects
`model_id`	string	No	—	Bedrock model ID (default: claude-3-haiku)
`system`	string	No	—	System prompt
`max_tokens`	integer	No	—	Max tokens to generate (default: 1024)

Field	Type	Required	Default	Description
`response_text`	string	Yes	—	The assistant's reply text
`model_id`	string	Yes	—	Bedrock model ID that was used
`input_tokens`	integer	Yes	—	Token count consumed by the input messages
`output_tokens`	integer	Yes	—	Token count generated in the response
`failure_reason`	string	No	—	—
`failure_type`	string	No	—	—
`failed_action`	string	No	—	—
`failed_at_state`	string	No	—	—
`failed_step`	string	No	—	—
`failed_layer`	string	No	—	—
`error`	string	No	—	—
`error_type`	string	No	—	—

State	Initial	Terminal	Success	Auto-advance	Description
`pending`	Yes	No	—	`execute`	—
`completed`	No	Yes	Yes	—	—
`failed`	No	Yes	No	—	—

From	Action	To	Description
`pending`	`execute`	`completed`	—
`*` (any state)	`fail`	`failed`	—

bash

POST /api/workflows/start
Content-Type: application/json

{
  "workflow_type": "aws.bedrock.chat",
  "initial_data": {
    "messages": "value"
  }
}