ChatGPT agent

Last updated Oct 17, 2026 Preview Inventory

AI summary

ChatGPT agents in your workspace are tracked in Rencore Governance through 28 properties organized in eight groups, including skills, connectors to external systems, run history, costs, ownership, and security risk. You can view which agents are published and can reach external systems, and take actions like unpublishing them or reviewing policies for agents that run without human intervention.

Agents published into the ChatGPT workspace, with their skills, connectors, run history and spend

Part of the ChatGPT Enterprise inventory.

In Rencore Governance

Scan settings

Default scanning interval	Allowed scanning intervals	Data removal during incremental scan
Daily	Daily, Weekly, Bi-Weekly, Monthly, Never, Initial	Yes

Properties

28 properties in 8 groups. Open a group to see its properties.

Name and identity 3

Property	Type	Automation placeholder
Definition Version	String	`{{ChatGPTAgent.DefinitionVersionId}}`
Display Name	String	`{{ChatGPTAgent.DisplayName}}`
ID	String	`{{ChatGPTAgent.Id}}`

Organization and location 2

Property	Type	Description	Automation placeholder
Workspace	ChatGPT Workspace		`{{ChatGPTAgent.Workspace}}`
Workspace Name	String		`{{ChatGPTAgent.WorkspaceName}}`

Activity and usage 2

Property	Type	Description	Automation placeholder
Last scan update	DateTime	Shows when this object was last updated in a scan. If an object is found during a scan but no property has been changed this date will not change.	`{{ChatGPTAgent.LastModifiedTime}}`
Status	String	Whether the agent is published and therefore live. Unpublish clears publication and deletes scheduled triggers without losing the draft. Allowed values: disabled, published, unpublished.	`{{ChatGPTAgent.Status}}`

AI and agents 6

Property	Type	Description	Automation placeholder
Connector Count	Int32		`{{ChatGPTAgent.ConnectorCount}}`
Connectors	String	Which external systems the agent can reach. A published agent with connectors can move data out of the workspace without a person in the loop.	`{{ChatGPTAgent.Connectors}}`
Model	String	The model the agent runs on.	`{{ChatGPTAgent.ModelSpec}}`
Skill Count	Int32		`{{ChatGPTAgent.SkillCount}}`
Skills	String	Which skills are live inside this agent.	`{{ChatGPTAgent.Skills}}`
Tools	String		`{{ChatGPTAgent.Tools}}`

Dates and scanning 3

Property	Type	Description	Automation placeholder
Created At	DateTime		`{{ChatGPTAgent.CreatedAt}}`
Initially Scanned	DateTime	Shows when this object was initially found in a scan.	`{{ChatGPTAgent.CreatedTime}}`
Updated At	DateTime		`{{ChatGPTAgent.UpdatedAt}}`

Property	Type	Automation placeholder
M365 Owner	User	`{{ChatGPTAgent.OwnerM365User}}`
M365 Owner Name	String	`{{ChatGPTAgent.OwnerM365UserName}}`
Owner	ChatGPT user	`{{ChatGPTAgent.OwnerUser}}`
Owner Email	String	`{{ChatGPTAgent.OwnerUserEmail}}`
Owner Name	String	`{{ChatGPTAgent.OwnerUserName}}`

Security and compliance 4

Property	Type	Description	Automation placeholder
Key	String	Internal.	`{{ChatGPTAgent.ApiKeyEncrypted}}`
Risk Score	Int32	Stores risk score	`{{ChatGPTAgent.RiskScore}}`
Risk Score Update	DateTime	Stores risk score update	`{{ChatGPTAgent.RiskScoreLastUpdate}}`
Risk Score Value	String	Stores risk score value like Low_Low	`{{ChatGPTAgent.RiskScoreValue}}`

Storage, licences and cost 3

Property	Type	Description	Automation placeholder
Costs Last 30 Days (USD)	Double		`{{ChatGPTAgent.CostsLast30DaysUsd}}`
Costs This Month (USD)	Double	Spend attributed to this agent through the COSTS log's agent identity.	`{{ChatGPTAgent.CostsThisMonthUsd}}`
Total Cost (USD)	Double		`{{ChatGPTAgent.TotalCostUsd}}`

Relations

Relation	Connector	Description
ChatGPT agent run	ChatGPT Enterprise	Runs of ChatGPT workspace agents, by trigger type and invoker
ChatGPT cost line	ChatGPT Enterprise	Daily ChatGPT cost lines by user, group, agent, product, surface and model
ChatGPT skill	ChatGPT Enterprise
ChatGPT user	ChatGPT Enterprise	Every user in the connected ChatGPT workspace, with role, seat, spend and the Entra ID join
ChatGPT Workspace	ChatGPT Enterprise	The connected ChatGPT Enterprise workspace, its granted scopes and its coverage gaps
User	Microsoft 365	All users registered in your tenant (internal, external)

Segments

Segment	Description
Published agents with connectors	Agents that are live and can reach an external system
Agents triggered without a human	Agents that ran from an API call, a webhook or a schedule in the last 30 days
Agents with spend	Workspace agents that have incurred cost this month

Actions

Delete Agent
Unpublish Agent

Policy	Severity	Description
Agent runs without a human and reaches external systems	Critical	Detects published agents that ran from a schedule, webhook or API call in the last 30 days and can reach a connector

Frequently asked questions

What is the ChatGPT agent object in Rencore Governance?

Agents published into the ChatGPT workspace, with their skills, connectors, run history and spend. ChatGPT agent is an object type in the ChatGPT Enterprise inventory of Rencore Governance.

Which properties does Rencore Governance collect for ChatGPT agent?

Rencore Governance collects 28 properties for ChatGPT agent, such as Key, Connector Count, Connectors and 25 more.

Which policies check ChatGPT agent objects?

1 built-in policy checks ChatGPT agent objects, such as Agent runs without a human and reaches external systems.

How often is ChatGPT agent scanned?

By default daily. You can change the scan interval of ChatGPT agent objects under Configure objects (daily, weekly, bi-weekly, monthly, never or initial (collected once)). The interval is a target, not a guarantee: a scan can run later than planned.

Which actions can I run on ChatGPT agent objects?

The built-in actions for ChatGPT agent objects are Delete Agent and Unpublish Agent.

Was this page helpful?