Noma Security Extends Agent Controls to Employee Endpoints

Noma Security announced in a press release that its platform can now discover and control AI agents, MCP servers and skills running on employee devices. The system connects through existing endpoint detection and response or mobile device management tools to identify assets using configuration files, skills directories, connector histories and running processes.

The platform assesses risks including exposed secrets, execution outside sandboxes and excessive permissions. Its access controls let security teams assign approved, review or blocked status, apply policies based on users and groups, and restrict specific tools and actions.

Noma's AI Detection and Response system monitors agent activity for prompt injection, sensitive data leaks, malicious intent, tool poisoning, scope violations and changes in behavior. Policies can monitor, alert, redirect or block activity, mask sensitive data, or send an action to a person for approval. Coverage includes Claude Code, Claude Cowork, Cursor, Codex, Windsurf, Kiro, Antigravity and OpenClaw.