CAIN VIGIL, explained for beginners

VIGIL checks every step an agent proposes and prunes destructive or exfiltrating ones.

What it is

Agents work step by step. A single bad step (deleting a database while 'summarizing a report', or posting a file to an outside server) can undo everything.

VIGIL scores each proposed step. It looks at the tool's name, its arguments, the intent and the justification. Destructive or exfiltration patterns are pruned; unjustified changes need a human; ordinary steps proceed. Each verdict is chained to the previous step.

An everyday comparison

A co-pilot who reads each instruction before it's carried out and refuses 'open the cargo door mid-flight'.

How it works, in 4 steps

1. The agent proposes a step.
2. VIGIL inspects tool, arguments and intent.
3. Safe: PROCEED. Risky: needs a human. Dangerous: PRUNE.
4. Each verdict is chained.

Try it now

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

See it on the homepage: CAIN VIGIL section →

Who it is for

Teams running multi-step agents with write access.

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.