Salt Security has announced new integrations with CrowdStrike’s Falcon platform to expand visibility into the activity of artificial intelligence (AI) agents within enterprises. The integration connects Salt Agentic API with Falcon Foundry, Falcon Next-Gen SIEM, and Falcon Firewall Management to identify agents, MCP servers, tools, APIs, and permissions, and to enable automated responses when detected behavior departs from established patterns.
Salt Security and CrowdStrike at a glance
- Salt integrates its API and AI agent platform with several CrowdStrike tools.
- The combination can identify agents, MCP servers, tools, APIs, and permissions.
- Findings can be correlated with identity, cloud, and device data in Falcon.
- CrowdStrike Firewall Management can trigger responses when an agent’s behavior changes.
- The collaboration between the two companies began in 2022 with a Falcon Fund investment in Salt Security.
The integration responds to a shift in how enterprises use artificial intelligence. Agents are no longer limited to generating text or answering questions: they can use credentials, connect to internal systems through MCP (Model Context Protocol) servers, use tools, and make API calls that trigger actions on users’ behalf.
For security teams, this introduces a new layer of oversight. Knowing which agents have been authorized doesn’t necessarily reveal which systems they’re using, what tools they have available, or what actions they perform during a session.
Salt Security and CrowdStrike have been working together since 2022, when Falcon Fund, CrowdStrike’s strategic investment vehicle, invested in Salt Security. Since then, the two companies had developed a certified Salt application for Falcon Foundry and a native integration with Falcon Next-Gen SIEM.
From agents to APIs and MCP servers
The new integration aims to extend that work to agent behavior. According to Salt Security, joint customers can use the certified application in Falcon Foundry to deploy Salt using the existing Falcon sensor, without adding extra gateways or proxies.
The system can discover the agents running within the environment, the MCP servers and tools through which they connect, the APIs those connections use, and the associated permissions.
One point the company highlights is the ability to also identify publicly exposed MCP servers and connections that haven’t gone through internal review processes.
The goal is to build a relationship between the agent and the resources it can access. An agent may have a legitimate function, but its operational risk also depends on the credentials it uses, the tools available to it, and the systems it can reach.
Salt contributes its platform specialized in APIs and agent-related activity, while CrowdStrike provides the rest of the security signals it already collects through Falcon.
Data on agent inventory, configuration, and behavior detections flows into Falcon Next-Gen SIEM, where it can be correlated with device, identity, and cloud environment information.
This correlation makes it possible to analyze agent activity alongside the rest of an organization’s security events, rather than keeping agent information in a separate tool.
Automatic responses when behavior changes
The second part of the integration concerns response. When an agent’s activity is found to depart from an established behavior pattern, customers can trigger actions through CrowdStrike Falcon Firewall Management.
Salt describes the flow as a chain that starts with discovering the agent and its connections, continues with analyzing permissions and behavior, and ends, when appropriate, with an action on network access.
The published information doesn’t specify that these responses run automatically in every scenario. They’re presented as capabilities customers can use within Falcon’s integrations and their own security policies.
This nuance matters because an agent’s activity can be legitimate even when it differs from its usual behavior. A change in the pattern of API calls, for example, doesn’t by itself mean there’s an intrusion. How useful these controls are depends on how each organization configures its policies, signals, and thresholds.
Salt and CrowdStrike’s proposal thus tries to address a specific problem in agentic system security: tracking the full path of an action from the agent to the tool, the API, and the final resource.
An agent can receive a valid instruction and, during execution, use a tool with excessive permissions or access an MCP connection that hadn’t been reviewed. The ability to observe that chain gives security teams more information to decide which access should be kept or restricted.
An integration that expands a relationship that started in 2022
The collaboration between the two companies didn’t start with this integration. In 2022, Falcon Fund made a strategic investment in Salt Security, and the two companies later brought Salt’s technology into CrowdStrike’s ecosystem.
The certified application for Falcon Foundry allows Salt to be deployed from existing Falcon infrastructure. The integration with Falcon Next-Gen SIEM then adds agent and API data to the security analysis system.
With the expansion announced in September 2026, the focus shifts specifically to AI agents and the connections they need to act within an organization.
The move also reflects the evolving security model around enterprise AI. As agents move from generating content to executing actions through tools and APIs, the surface that needs oversight is no longer limited to the AI model itself. It also includes credentials, permissions, MCP servers, APIs, and the systems the agent can access.
Salt Security has framed the integration as a way to answer three questions from a single environment: which agents are running, what they can access, and whether their current behavior matches those permissions. It’s a similar goal to the one behind Portnox’s network “kill switch” for risky AI agents, and it builds on CrowdStrike’s own push into AI agents with Falcon IQ.
Frequently Asked Questions
What does Salt Security integrate with CrowdStrike?
Salt integrates its Agentic API platform with CrowdStrike Falcon Foundry, Falcon Next-Gen SIEM, and Falcon Firewall Management to provide visibility into AI agents, their connections, APIs, and permissions.
What information can customers discover?
The integration can identify agents, MCP servers, tools, APIs in use, and associated permissions. It also covers connections and MCP servers that may be exposed to the internet.
What happens when an agent changes its behavior?
Findings can reach Falcon Next-Gen SIEM and, when configured conditions are met, can trigger responses through Falcon Firewall Management.
Since when have Salt Security and CrowdStrike been working together?
The companies have worked together since 2022, when Falcon Fund, CrowdStrike’s strategic investment vehicle, invested in Salt Security.
