This page provides a reference for AWS Bedrock actions available in the workflow automation actions catalog. Use these actions to invoke Bedrock agents as part of your workflow definitions.
Prerequisites
Before using AWS Bedrock actions in workflow automation, ensure you have:
- An AWS account with appropriate permissions.
- AWS credentials configured (IAM user credentials, IAM role ARN, or session credentials). At least one credential option must be provided — IAM role takes precedence over static access keys.
- The necessary IAM permissions for Bedrock operations (
bedrock:InvokeAgent,bedrock:InvokeModel).
See Set up AWS credentials for information on how to create IAM users and IAM roles, and set up static and session AWS credentials for integration with workflow automation AWS actions.
Invoke a Bedrock agent
The action identifier is aws.bedrock.invokeAgent.
Invokes an AWS Bedrock agent with a prompt and returns the agent's response.
The following table describes the input fields for this action.
Input | Type | Description | Example |
|---|---|---|---|
| String | Optional. IAM role ARN to assume for AWS API calls. |
|
| String | Optional. AWS access key ID. Pass as a secret. |
|
| String | Optional. AWS secret access key. Pass as a secret. |
|
| String | Optional. Temporary session token for STS authentication. Pass as a secret. |
|
| String | Required. AWS region where the Bedrock agent is located. |
|
| String | Required. The unique identifier of the Bedrock agent. |
|
| String | Required. The alias identifier for the Bedrock agent. |
|
| String | Required. The prompt to send to the agent. |
|
| String | Optional. Session identifier to maintain conversation context across multiple invocations. |
|
| Map | Optional. Additional boto3 fields not otherwise modeled. |
|
| List | Optional. JQ selectors to extract specific fields from the action output. |
|
The following table describes the output fields for this action.
Output | Type | Example |
|---|---|---|
| Object |
|
| Boolean |
|
| String |
|