MCP Attestor, explained for beginners

MCP Attestor doesn't just read an AI tool's description; it runs the tool in a sandbox to prove it does what it says.

What it is

Reading a tool's description catches tools that say something suspicious. It cannot catch a tool that says the right thing and does the wrong thing, like a 'search' tool that ignores your query and always returns the same canned answer.

MCP Attestor does both. The scan reads the description and code. The probe runs the tool in a sealed sandbox with tests (does a different query give a different answer?). If a hardened version is produced, it must pass the same tests; the result is signed into a chained record.

An everyday comparison

A car inspection that reads the paperwork and then actually test-drives the car.

How it works, in 4 steps

1. Scan: read the tool's description and code.
2. Probe: run it in a sandbox with behaviour tests.
3. Harden: the fixed version must pass the same tests.
4. Sign the result into a chained record.

Try it now

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

See it on the homepage: MCP Attestor section →

Who it is for

Companies relying on MCP tools for real work, and tool publishers who want to prove their tools behave.

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.