Linx Security adds identity controls for managed and shadow AI agents

Linx Security has introduced four capabilities designed to bring managed and unmanaged AI agents under identity control. The tools cover shadow agent discovery, runtime enforcement, autonomous identity operations and plain-English workflow creation.
- Shadow AI Agent Detection finds unsanctioned agents across endpoints, pipelines, SaaS and cloud.
- Agent Runtime Control evaluates tool calls through an existing gateway or the Linx MCP Gateway.
- Autopilots support scoped, approval-based identity operations.
- Prompt-to-Workflow converts plain-English descriptions into tested workflows.
- Linx says customers govern more than 25,000 AI agent identities in production.
Four capabilities extend identity governance
Shadow AI Agent Detection identifies agents operating outside sanctioned platforms, including those running on endpoints, in CI/CD pipelines, and across SaaS and cloud environments. Each finding is added to the Linx Identity Graph with information about its owner, potential exploitation and possible remediation.
Agent Runtime Control allows organizations to use the AI gateway they already operate, with Linx acting as its decision point through the AuthZEN open standard. Organizations without an existing gateway can use the Linx MCP Gateway. Linx says every agent tool call can be allowed or blocked according to the identity behind it, its risk and the action it is attempting.
The company also introduced Autopilots for any use case. Identity teams can assign an agent a mission in plain language, define its scope and choose which actions it can take independently, which require approval and which it cannot take. Teams can review and dry-run an Autopilot's plan before activating it.
Prompt-to-Workflow lets teams describe a process in plain English. Linx gives the example of deactivating an agent when its owner leaves. The platform builds the workflow, displays the underlying code and supports a dry run before it goes live.
Linx frames agent security as an identity problem
Linx says AI agents create challenges that traditional controls built around people may not handle well. Agents can be ephemeral, spawned, renamed, modified and reactivated with little trace. They can also act autonomously across company tools and data, while shadow agents may run on personal accounts, in pipelines and in AI applications outside sanctioned platforms.
The company’s approach is based on placing every sanctioned or unsanctioned agent on one Identity Graph alongside human and non-human identities. Linx says the graph records ownership and the access an agent carries. That context is then used to assess risk, set policy for tool calls and access changes, and enforce decisions in the gateway through which agents operate.
“AI agents are fundamentally changing the identity security problem,” Israel Duanis, co-founder and CEO of Linx Security, said in the announcement. “Enterprises need to know not just who has access, but what every identity is connected to, what it's doing and why.”
The company says agent governance is already expanding
Linx said its customers already govern more than 25,000 AI agent identities in production, while governed agent identities grew 450x in six months. The company also cited a Gartner prediction that through 2028, over 50% of enterprise AI initiatives will halt because of unresolved agentic identity challenges.
The capabilities are intended to extend identity governance from deciding who receives access to controlling what AI agents do after they receive it. Linx says the tools are designed for identity and security teams that need visibility into unmanaged agents, policy enforcement at runtime and governed automation without replacing existing gateway infrastructure.
Linx will demonstrate the capabilities at the Gartner Identity & Access Management Summit in Las Vegas from December 7, 9, 2026.

