Salt Security

Salt Security secures the full agentic AI stack: the LLMs, MCP servers, and APIs that AI agents rely on to take action. Founded in Israel in 2016 and headquartered in Palo Alto, Salt is a pioneer in API security and now extends that foundation to protect the agentic ecosystem that enterprises are building. The problem: AI agents act through APIs. Every tool call an agent makes ends in an API request, so risk moves inside the application layer, where edge controls, WAFs, and gateways have limited visibility. Organizations often cannot see which agents and MCP servers exist, what they are permitted to do, or when an agent takes an action nobody intended or authorized. The Salt Agentic Security Platform addresses this with two core capabilities. Agentic Security Posture Management (AG-SPM) continuously discovers the agentic lifecycle from code to runtime. It builds an Agentic Security Graph that maps relationships between LLMs, MCP servers, agents, and the APIs they call. Security teams can find shadow MCP servers, identify over-permissioned agents, and understand sensitive data exposure and risky configurations before they are exploited. Agentic Detection and Response (AG-DR) provides runtime protection. It builds agent-aware behavioral baselines, attributes activity to individual agent identities, and detects threats such as mass data extraction, unauthorized tool use, and abuse of business logic. It can interrupt machine-speed attacks in real time. AG-DR also includes native AI Detection and Response (AI-DR), which protects LLMs against direct and indirect prompt injection, jailbreaks, and unsafe model behavior, and links model-level attacks to downstream MCP and API activity.