MCPGate Guard Stack, explained for beginners

The Guard Stack checks every tool call an AI agent makes; ordinary work goes through and each attack is stopped by a different guard.

What it is

AI agents use tools through MCP (the Model Context Protocol): look up a customer, store a note, run a command, send a payment. The guard stack checks every one of those calls before it reaches the tool.

In the demo, one agent does a normal day's work (allowed), a payment is held for a human, then five attacks are tried: reading a vault outside its skills, paying through a delegation loop, a forged 'already approved' memory, a typosquatted package install, and a 60 MB export. Each is refused, by a different guard.

An everyday comparison

Airport security with several checkpoints: boarding pass, bag scan, passport, and a guard at the gate.

How it works, in 4 steps

1. The agent calls a tool.
2. Each guard in the stack checks the call.
3. All agree: the tool runs.
4. Any refuses: the call never reaches the tool.

Try it now

Open the six live demo tools → Every button sends a real request to the live product and shows the real answer.

See it on the homepage: MCPGate Guard Stack section →

Who it is for

Teams whose agents use MCP tools on real systems.

FIRME AI CORPORATION · CAIN-42 is a Byzantine trust fabric. Numbers marked "our own check" were run by us, not by an independent party. Nothing on these pages is a certification.