Microsoft Defender for Endpoint
Endpoint protectionLocal Agent Discovery Overview
In brief
The documentation now distinguishes process-based discovery from VS Code extension collection on Windows and macOS. It also states that MCP server configuration discovery requires antivirus active mode and real-time protection, and removes the macOS preview note.
What Defender admins need to know
Ensure onboarded clients are running and reporting. Enable antivirus active mode and real-time protection if MCP server configuration discovery is required.
Summaries are generated from the documentation change itself.
Documentation change
View on Microsoft LearnThe comparison below shows only the changed extract. Use the full-page view for complete context.
Local AI agents run with user-level permissions and can access files, tools, and services on the devices where they operate. Without visibility into which agents are running and what they can reach, security teams can't assess exposure, enforce governance, or respond to agent-related incidents.
Microsoft Defender can discoverdiscovers local AI agents and their configured MCP servers after it observes agent activity on onboarded devices, thendevices. It surfaces themthese agents in the Microsoft Defender portal. This gives security teams a centralized view of AI agents usedpresent in the organization.
Local AI agent discovery on endpoints
Discovery depends on the agent type. For process-based discovery, Defender lists a local AIidentifies an agent afterwhen it observes the agent activity on an onboarded device. Installing an agent's process running. The user doesn't by itself guarantee thatneed to submit a prompt or have permission to use the agent appears's service. An inventory entry isn't proof of active use or successful access to an AI service.
On Windows and macOS, the onboarded Defender for Endpoint client also periodically collects installed VS Code extension data on the endpoint. The client must be running and reporting. Agent inventory discovery through this collection doesn't require antivirus active mode or the extension to run. See the entries marked with footnote 1 and their preview status in the local AI agent inventory. An installed agent might not appear until Defender observes agent activity.support matrix.
MCP server configuration discovery still requires antivirus active mode and real-time protection. For details and requirements, see the local AI agent discovery prerequisites.
Local AI agent discovery isn't a complete inventory of installed software. It provides security context for agents that Defender observes on endpoints, including the device and account associated with the activity.agent. To review known software installed on devices, use Software inventory in Microsoft Defender Vulnerability Management.
When Defender identifies a local AI agent, the agent is displayed as a discoverable asset in the Microsoft Defender portal with visibility into:
- Exposure map: Visual relationships between local AI agents, devices, identities, and the resources those identities can access, to help assess potential impact.
- Advanced hunting: Hunting for discovery data using Kusto Query Language (KQL) to investigate local AI agents and the resources they can access based on the permissions of the user running them.
Note
Local AI agent discovery on macOS is in preview.
Supported local AI agents and MCP server configurations
Defender defines an agent as a combination of a user, a device, and an agent type. For example, if Claude Code runs in 15 different project folders on the same device for the same user, it appears as a single agent entry in the inventory.
@@ -14,7 +14,7 @@ ms.author: lwainstein ms.service: defender-endpoint ms.topic: overview ms.custom: msecd-doc-authoring-1030-ms.date: 2026-10-07T00:00:00.0000000Z+ms.date: 2026-11-02T00:00:00.0000000Z ai-usage: ai-assisted locale: en-us document_id: d76258bb-b8c4-0e21-0c6a-94d5623ec4e6@@ -44,7 +44,7 @@ platformId: 5fbe46e6-1635-6e3d-c6d8-da353b0d5708 Local AI agents run with user-level permissions and can access files, tools, and services on the devices where they operate. Without visibility into which agents are running and what they can reach, security teams can't assess exposure, enforce governance, or respond to agent-related incidents.
-Microsoft Defender can discover local AI agents and their configured MCP servers after it observes agent activity on onboarded devices, then surfaces them in the Microsoft Defender portal. This gives security teams a centralized view of AI agents used in the organization.
+Microsoft Defender discovers local AI agents and their configured MCP servers on onboarded devices. It surfaces these agents in the Microsoft Defender portal. This gives security teams a centralized view of AI agents present in the organization.
[](media/local-agent-discovery-overview/discovery-overview.png#lightbox)
@@ -56,9 +56,13 @@ Defender also provides **AI agent runtime protection** for local agents. When en
## Local AI agent discovery on endpoints
-Defender lists a local AI agent after it observes agent activity on an onboarded device. Installing an agent doesn't by itself guarantee that the agent appears in the local AI agent inventory. An installed agent might not appear until Defender observes agent activity.
+Discovery depends on the agent type. For process-based discovery, Defender identifies an agent when it observes the agent's process running. The user doesn't need to submit a prompt or have permission to use the agent's service. An inventory entry isn't proof of active use or successful access to an AI service.
-Local AI agent discovery isn't a complete inventory of installed software. It provides security context for agents that Defender observes on endpoints, including the device and account associated with the activity. To review known software installed on devices, use [Software inventory in Microsoft Defender Vulnerability Management](/en-us/defender-vulnerability-management/tvm-software-inventory).
+On Windows and macOS, the onboarded Defender for Endpoint client also periodically collects installed VS Code extension data on the endpoint. The client must be running and reporting. Agent inventory discovery through this collection doesn't require antivirus active mode or the extension to run. See the entries marked with footnote 1 and their preview status in the [support matrix](ai-agent-support-matrix#ai-agent-discovery-support).
+
+MCP server configuration discovery still requires antivirus active mode and real-time protection. For details and requirements, see the [local AI agent discovery prerequisites](discover-local-ai-agents#prerequisites).
+
+Local AI agent discovery isn't a complete inventory of installed software. It provides security context for agents that Defender observes on endpoints, including the device and account associated with the agent. To review known software installed on devices, use [Software inventory in Microsoft Defender Vulnerability Management](/en-us/defender-vulnerability-management/tvm-software-inventory).
When Defender identifies a local AI agent, the agent is displayed as a discoverable asset in the Microsoft Defender portal with visibility into:
@@ -66,10 +70,6 @@ When Defender identifies a local AI agent, the agent is displayed as a discovera - **Exposure map**: Visual relationships between local AI agents, devices, identities, and the resources those identities can access, to help assess potential impact.
- **Advanced hunting**: Hunting for discovery data using Kusto Query Language (KQL) to investigate local AI agents and the resources they can access based on the permissions of the user running them.
-Note
-
-Local AI agent discovery on macOS is in preview.
-
## Supported local AI agents and MCP server configurations
Defender defines an agent as a combination of a user, a device, and an agent type. For example, if Claude Code runs in 15 different project folders on the same device for the same user, it appears as a single agent entry in the inventory.