Collected AI Attributes
AI Automations discover metadata from connected platforms and map it to system attributes, which appear on the details screen of the matching inventory record.
For more information about AI Automations, see the following:
Common Attributes
Common attributes apply to both agents and models. Each is listed with the platforms that populate it for each entity type.
|
Attribute |
Description |
Agent platforms |
Model platforms |
|---|---|---|---|
|
Creation Source |
How the record was created. For automation-discovered records, the value is Automation. |
All supported agent automations |
AWS Bedrock, AWS SageMaker, Azure Foundry, GCP Vertex, Databricks |
|
Description |
Description of the agent or model and its purpose. |
AWS Bedrock, Amazon Bedrock AgentCore |
AWS Bedrock, AWS SageMaker, Azure Foundry, GCP Vertex, Databricks |
|
Full Resource Name |
Fully qualified platform resource identifier. For AgentCore agents, populated from Full Source Name. |
All supported agent automations |
AWS Bedrock, Azure Foundry |
|
Scope |
Platform scope or path for the discovered asset. The exact path structure varies by platform. |
All supported agent automations |
AWS Bedrock, Azure Foundry |
|
Stage |
Lifecycle stage of the agent or model. Uses the platform-supported lifecycle value where available. |
All supported agent automations |
AWS SageMaker, Databricks |
Agent Attributes
These attributes are collected for agents discovered through AI Automations.
|
Attribute |
Supported platforms |
Description |
Notes |
|---|---|---|---|
|
Agent ARN / Resource ID |
All supported agent automations |
Platform resource identifier for the agent. |
Bedrock uses an ARN. Azure Foundry uses the agent name. |
|
Agent ID |
All supported agent automations |
Platform-specific unique identifier for the agent. |
Shown in AgentCore as AgentId. |
|
Agent Name |
All supported agent automations |
Display name of the agent. |
Shown in AgentCore as Name. |
|
Agent Source |
All supported agent automations |
Source platform from which the agent was discovered. |
AgentCore is reported as a Bedrock source value. |
|
Agent Version |
All supported agent automations |
Version of the agent. |
Shown in AgentCore as AI Agent Version. |
|
Agent Type |
AWS Bedrock |
Classification of the agent. |
|
|
Agentic Compound System |
All supported agent automations |
Indicates whether the discovered record is an agentic compound system. |
Always Yes for discovered agents. |
|
Autonomy Level |
AWS Bedrock |
Degree of agent autonomy. |
|
|
Created By |
Amazon Bedrock |
User or principal that created the agent. |
|
|
Default Model |
Azure Foundry |
Default model associated with the agent. |
|
|
Hosting Environment |
All supported agent automations |
Platform hosting the agent. |
Example: Amazon Web Services (AWS) Bedrock. |
|
Instructions |
AWS Bedrock |
Instructions that define how the agent behaves. |
|
|
Is Deployed |
All supported agent automations |
Indicates whether the agent is deployed. |
Based on the platform’s deployment or publish state. |
|
Last Updated By |
Amazon Bedrock |
User or principal that last updated the agent. |
|
|
Linked Identity |
Amazon Bedrock AgentCore |
Identity linked to the agent. Includes the credential type, linked identity name, workload linked identity, linked identity created date, and linked identity last updated date. |
|
|
Operational Status |
All supported agent automations |
Current operational state of the agent. |
AgentCore Operational Status values are normalized to this attribute. |
|
Organization |
Amazon Bedrock AgentCore |
Organization or policy domain associated with the agent. |
|
|
Region |
All supported agent automations |
Cloud region where the agent runs. |
Shown in AgentCore as Agent Region. |
|
Underlying Model |
All supported agent automations |
Foundation or backing model used by the agent. |
|
|
Version Aliases |
AWS Bedrock |
Version-to-alias mapping for the agent. |
Populated only when aliases exist. |
Model Attributes
These attributes are collected for models discovered through AI Automations.
|
Attribute |
Supported platforms |
Description |
|---|---|---|
|
Created Date |
AWS Bedrock, AWS SageMaker, Azure Foundry, GCP Vertex, Databricks |
Date the model record was created. |
|
Deployment ID |
Azure Foundry |
Deployment identifier for the model. |
|
Endpoint URL |
Azure Foundry |
Endpoint address for the model. |
|
Framework |
AWS Bedrock, AWS SageMaker, Azure Foundry, GCP Vertex, Databricks |
Machine learning framework used by the model. |
|
Generative AI |
AWS Bedrock, Databricks |
Indicates whether the model is generative AI. |
|
Hosting Provider |
AWS Bedrock, AWS SageMaker, Azure Foundry, GCP Vertex, Databricks |
Provider hosting the model. |
|
Mode |
AWS Bedrock, Databricks |
Operating mode for the model. |
|
Model ID |
AWS Bedrock, AWS SageMaker, Azure Foundry, GCP Vertex, Databricks |
Platform-specific unique identifier for the model. |
|
Model Name |
AWS Bedrock, AWS SageMaker, Azure Foundry, GCP Vertex, Databricks |
Display name of the model. |
|
Model Source |
AWS Bedrock, AWS SageMaker, Azure Foundry, GCP Vertex, Databricks |
Source platform from which the model was discovered. |
|
Model Version |
AWS Bedrock, AWS SageMaker, Azure Foundry, GCP Vertex, Databricks |
Version of the model. |
|
Programming Language |
AWS Bedrock, Databricks |
Programming language used to implement the model. |
|
Type of Learning |
AWS Bedrock, Databricks |
Learning method used to train the model. |
Guardrail Attributes
This section lists the supported guardrail attributes for agents and models.
Azure Foundry models use the same guardrail attributes as Azure Foundry agents, but Azure Foundry and Amazon Bedrock do not support the exact same guardrail types.
|
Display label |
System field label |
Supported systems |
|---|---|---|
|
Automated Reasoning |
Has Automated Reasoning Guardrail |
AWS Bedrock |
|
Content Filtering — Input |
Has Content Filtering Guardrail - Input |
AWS Bedrock, Azure Foundry |
|
Content Filtering — Output |
Has Content Filtering Guardrail - Output |
AWS Bedrock, Azure Foundry |
|
Contextual Grounding |
Has Contextual Grounding Guardrail |
AWS Bedrock, Azure Foundry |
|
Denied Topics — Input |
Has Denied Topics Guardrail - Input |
AWS Bedrock |
|
Denied Topics — Output |
Has Denied Topics Guardrail - Output |
AWS Bedrock |
|
Prompt Attack Protection |
Has Prompt Attack Guardrail |
AWS Bedrock, Azure Foundry |
|
Protected Material — Code |
Has Protected Material - Code Guardrail |
Azure Foundry |
|
Protected Material — Text |
Has Protected Material - Text Guardrail |
Azure Foundry |
|
Sensitive Info Filters — Input |
Has Sensitive Information Filters Guardrail - Input |
AWS Bedrock, Azure Foundry |
|
Sensitive Info Filters — Output |
Has Sensitive Information Filters Guardrail - Output |
AWS Bedrock, Azure Foundry |
|
Word Filters — Input |
Has Word Filters Guardrail - Input |
AWS Bedrock, Azure Foundry |
|
Word Filters — Output |
Has Word Filters Guardrail - Output |
AWS Bedrock, Azure Foundry |
Dataset Attributes
Dataset attributes are currently supported for Databricks only.
|
Attribute |
Supported system |
Description |
|---|---|---|
|
Catalog Metadata |
Databricks |
Structured catalog metadata for the dataset. |
|
Dataset Version |
Databricks |
Dataset version value. |
|
Full Name |
Databricks |
Full dataset name. |
|
Unique Identifier |
Databricks |
Platform-specific unique identifier for the dataset. |