Evoke Security is a security platform designed to provide visibility, governance, and threat detection for agentic AI workforces — environments where autonomous AI agents operate alongside human employees. The platform addresses the security risks introduced by AI agents that have broad access to organizational resources such as email, messaging, customer data, internal databases, and source code. It treats each agent as a potential single point of failure and provides security teams with the tools to monitor and control agent behavior. Evoke Security operates across three deployment contexts: - Local agents (via an endpoint agent) - Production agents (via SDK and proxy API) - SaaS agents (via a browser extension) Core capabilities are organized into three functional areas: Inventory & Governance: Automatically discovers all connected AI agents, models, tools, and data sources within an organization. Maps agent skills and MCP (Model Context Protocol) connections to eliminate shadow AI. AI Security Posture Management (AI-SPM) & Threat Modeling: Identifies malicious agent skills and MCPs, maps attack paths and toxic data flows, and remediates over-permissioned agents to reduce blast radius before a compromise occurs. Detection, Prevention & Response: Monitors all prompts, tool calls, and agent responses in real time. Enforces per-agent security policies and blocks unauthorized actions before damage occurs.