Connect
Add a capability we have never heard of.
Your warehouse system, your pricing engine, your internal service. Describe it once — name, inputs, output, authentication, risk level — and it becomes a registry entry that behaves exactly like every built-in one: permissioned, approvable, audited and revocable.
Defined once, governed like the rest
A tool you wrote gets the same permission checks, approval behaviour, rate limits and audit records as a built-in capability. There is no second, weaker path for custom code.
You choose how narrow it is
One endpoint can become one tool or six. Six narrow tools are almost always the better choice, because each can be granted to a different agent and refused to the others.
You set its risk level, and we hold you to it
A tool you declare as low risk but which writes to a ledger will be caught in review. Risk is declared by you and checked against what the tool actually does.
Credentials stay in the vault
Your service's authentication is held as a secret and injected at call time. It is never placed in an agent's context, which means it cannot be repeated back by a model or read out of a log.
Tools
The rest of this area.
Named capabilities with defined inputs, outputs and risk.
The tool registry
Every capability an agent could be given, in one list.
What a tool declares
Inputs, outputs, risk, permission, approval behaviour.
Risk levels
Four levels, and the policies that attach to them.
Your business as a toolbox
Connected systems become capabilities, not open access.
Adding external systems
The same architecture, more tools. No redesign.
Give AI a job.
Build an agent, choose what it can access, define what it may do, and deploy it to work safely across your business.
No code required · Full SDKs and API · Stop any agent instantly · [email protected]