AWS::BedrockAgentCore::Harness

Specifies a harness for Amazon Bedrock AgentCore. A harness is a managed agent loop that lets you define and invoke AI agents with a single API call. Specify a model, system prompt, and tools inline — the harness handles orchestration, tool execution, memory management, and response generation. Each session runs in an isolated microVM with filesystem and shell access, supporting use cases like code generation, data analysis, and deep research.

For more information, see Build agents with Amazon Bedrock AgentCore Harness.

See the Properties section below for descriptions of both the required and optional properties.

Syntax

To declare this entity in your CloudFormation template, use the following syntax:

JSON


{
  "Type" : "AWS::BedrockAgentCore::Harness",
  "Properties" : {
      "AllowedTools" : [ String, ... ],
      "AuthorizerConfiguration" : AuthorizerConfiguration,
      "Environment" : HarnessEnvironmentProvider,
      "EnvironmentArtifact" : HarnessEnvironmentArtifact,
      "EnvironmentVariables" : {Key: Value, ...},
      "ExecutionRoleArn" : String,
      "HarnessName" : String,
      "MaxIterations" : Integer,
      "MaxTokens" : Integer,
      "Memory" : HarnessMemoryConfiguration,
      "Model" : HarnessModelConfiguration,
      "Skills" : [ HarnessSkill, ... ],
      "SystemPrompt" : [ HarnessSystemContentBlock, ... ],
      "Tags" : [ Tag, ... ],
      "TimeoutSeconds" : Integer,
      "Tools" : [ HarnessTool, ... ],
      "Truncation" : HarnessTruncationConfiguration
    }
}

YAML


Type: AWS::BedrockAgentCore::Harness
Properties:
  AllowedTools: 
    - String
  AuthorizerConfiguration: 
    AuthorizerConfiguration
  Environment: 
    HarnessEnvironmentProvider
  EnvironmentArtifact: 
    HarnessEnvironmentArtifact
  EnvironmentVariables: 
    Key: Value
  ExecutionRoleArn: String
  HarnessName: String
  MaxIterations: Integer
  MaxTokens: Integer
  Memory: 
    HarnessMemoryConfiguration
  Model: 
    HarnessModelConfiguration
  Skills: 
    - HarnessSkill
  SystemPrompt: 
    - HarnessSystemContentBlock
  Tags: 
    - Tag
  TimeoutSeconds: Integer
  Tools: 
    - HarnessTool
  Truncation: 
    HarnessTruncationConfiguration

Properties

AllowedTools

The allowed tools of the harness. All tools are allowed by default.

Required: No

Type: Array of String

Minimum: 1

Maximum: 64

Update requires: No interruption

AuthorizerConfiguration

Represents inbound authorization configuration options used to authenticate incoming requests.

Required: No

Type: AuthorizerConfiguration

Update requires: No interruption

Environment

The compute environment on which the Harness runs.

Required: No

Type: HarnessEnvironmentProvider

Update requires: No interruption

EnvironmentArtifact

The environment artifact (e.g., container) in which the Harness operates.

Required: No

Type: HarnessEnvironmentArtifact

Update requires: No interruption

EnvironmentVariables

Environment variables exposed in the environment in which the harness operates.

Required: No

Type: Object of String

Pattern: ^[\s\S]*$

Minimum: 0

Maximum: 5000

Update requires: No interruption

ExecutionRoleArn

IAM role the harness assumes when running.

Required: Yes

Type: String

Pattern: ^arn:aws(-[^:]+)?:iam::[0-9]{12}:role/.+$

Minimum: 1

Maximum: 2048

Update requires: No interruption

HarnessName

The name of the harness.

Required: Yes

Type: String

Pattern: ^[a-zA-Z][a-zA-Z0-9_]{0,39}$

Update requires: Replacement

MaxIterations

The maximum number of iterations in the agent loop allowed before exiting per invocation.

Required: No

Type: Integer

Update requires: No interruption

MaxTokens

The maximum total number of output tokens the agent can generate across all model calls within a single invocation.

Required: No

Type: Integer

Update requires: No interruption

Memory

AgentCore Memory instance configuration for short and long term memory.

Required: No

Type: HarnessMemoryConfiguration

Update requires: No interruption

Model

The configuration of the default model used by the Harness.

Required: Yes

Type: HarnessModelConfiguration

Update requires: No interruption

Skills

The skills of the harness.

Required: No

Type: Array of HarnessSkill

Update requires: No interruption

SystemPrompt

The system prompt of the harness.

Required: No

Type: Array of HarnessSystemContentBlock

Update requires: No interruption

Tags

Tags to apply to the harness resource.

Required: No

Type: Array of Tag

Maximum: 50

Update requires: No interruption

TimeoutSeconds

The maximum duration per invocation.

Required: No

Type: Integer

Update requires: No interruption

Tools

The tools of the harness.

Required: No

Type: Array of HarnessTool

Update requires: No interruption

Truncation

Configuration for truncating model context.

Required: No

Type: HarnessTruncationConfiguration

Update requires: No interruption

Return values

Ref

Fn::GetAtt

The Fn::GetAtt intrinsic function returns a value for a specified attribute of this type. The following are the available attributes and sample return values.

For more information about using the Fn::GetAtt intrinsic function, see Fn::GetAtt.

Arn: The ARN of the harness.
CreatedAt: The createdAt time of the harness.
Environment.AgentCoreRuntimeEnvironment.AgentRuntimeArn: The ARN of the underlying AgentCore Runtime.
Environment.AgentCoreRuntimeEnvironment.AgentRuntimeId: The ID of the underlying AgentCore Runtime.
Environment.AgentCoreRuntimeEnvironment.AgentRuntimeName: The name of the underlying AgentCore Runtime.
HarnessId: The ID of the harness.
Memory.ManagedMemoryConfiguration.Arn: The ARN of the harness.
Status: The status of the harness.
UpdatedAt: The updatedAt time of the harness.
Version: The version of the harness. Incremented on every successful UpdateHarness.

Document Conventions

ToolSchema

AuthorizerConfiguration