What is the Salt Agentic Security Platform?

Salt secures AI agents and the LLMs, MCP servers and APIs they act through. It gives you breadth across every agent, MCP server and API, wherever it runs, and depth into the exposure, source code, configuration and runtime behavior behind each one.

How is agentic security different from API security?

Agentic security protects what AI agents do, not just the APIs they call. Every action an agent takes ultimately becomes an API call, often through an MCP server, so Salt follows the whole path from agent to model to MCP server to API instead of judging one step in isolation.

What are AG-SPM, AI-DR and AG-DR?

They are Salt’s three capabilities. AG-SPM (Agentic Security Posture Management) finds agents, MCP servers and tools and flags risk before they run. AI-DR (AI Detection and Response) watches prompts and model responses. AG-DR (Agentic Detection and Response) detects and blocks harmful agent actions in real time.

Can Salt find shadow AI agents and MCP servers?

Yes. Salt discovers every agent, MCP server and tool, including shadow MCP servers nobody registered, and flags exposed endpoints, hardcoded secrets and over-scoped access.

Which compliance frameworks does Salt map agent posture to?

Salt maps agent posture to the EU AI Act, NIST and SOC 2, so you can show auditors how your agents, MCP servers and APIs are governed.

Does Salt work with the tools we already use?

Yes. Salt has 40+ integrations across cloud, gateways, code and the SOC, including AWS, Azure, Apigee, Kong, Kubernetes, Docker, Splunk, Cloudflare and Wiz.

The Agentic Security Platform See every agent. Secure every layer.

Salt gives you breadth across every agent, LLM, MCP server and API, wherever it runs, and depth into the exposure, code, configuration and runtime behind each one.

Book a demo

Finance MCPInside the MCP server

Exposure

reachable from: internal onlyInternet-facing

Source code

apiKey: process.env.KEYHard-coded key

Configuration

scope: records:readOver-scoped

Runtime behavior

GET /records/1041 200Enumerating records

Try it yourself · Inside the Salt Console

Click through a live environment.

This is the Salt console on one real environment. Watch the short tour, then take over: click any agent, MCP server, technology or app to follow its path all the way down to the risk factors behind it.

1Pick an agent
2Follow its path
3Open the risk factors

Guided tour · click anywhere to take over

Agentic Coverage Live console

Agentic

Insights

Insights:

Risk Score

Posture Gaps

Layers

Layers:

Technologies

Applications

Fit to screen

Agents

weather-assistant-agent

bedrock-agentcore.us-east-1.amaz…/runtimes/…

agent-quick-start-2…

bedrock-agentcore.us-east-1.amaz…/runtimes/…

MCPs & Toolkit

Other

Toolkit

6.5

MCP

store.duolingo.com/api/mcp

6.5

MCP

store-eu.duolingo.com/api/mcp

4.7

aws

gateway-quick-start-4b0eec…

execute-api.us-east-1.amazonaws.com

Technologies

6.5

4.7

Applications

6.5

Capabilities

One platform · Three capabilities

Before deployment. At the prompt. At the action.

Point tools see one stage. Salt carries context from each stage into the next, along the same path an agent's request takes.

Salt found MCP servers we didn't know existed, including two exposed to the internet. That changed how we approve every new agent.

Head of Application Security · Healthcare technology company

49.8%

of organizations confirmed or suspected an AI agent took an action they did not intend or authorize in the past year.

Salt Security, 1H 2026 State of AI and API Security report

Built for the enterprise you have

Keep the stack you have. Salt connects to all of it.

Start with agentless discovery, then add protection through the gateways, WAFs and SOC tools you already run. Nothing to rip out, reroute or rewrite.

40+ technology integrations across cloud, gateways, edge, code and the SOC.

See all integrations

FAQ

Frequently asked questions about the Agentic Security Platform.

Your agents are already acting.
See your own graph.

Find the agents and MCP servers you didn't know about, see what they can reach, and secure them at every stage.

Get an Agentic Security assessment Book a demo