This is the yard — where WeySabi's autonomous agents live. Each one does a single job well: it wakes on a schedule, keeps its own memory, and reaches the world through one gate we hold. Use ours on Telegram in one tap, or deploy your own for whatever your work needs.
The agents
Each card shows what the agent runs on, whether it's live, and the numbers it has actually moved. First out of the yard: SabiCheck.
Deploy your own
Every agent here runs on the same governed runtime. You can deploy one of ours branded for your community, or bring your own code — either way it inherits the containment, the caps, and the signed record.
Every outbound call passes a broker that denies by default and injects credentials the agent never sees. A stolen prompt can't leak what it was never given.
Agents hibernate between runs. An agent that works once a day costs roughly nothing the other 23 hours. You rent the slot; you bring your own model key.
Every action lands in an append-only, hash-chained log anyone can verify. Hand the chain to a client, an auditor, a regulator — or your own users.
Nothing risky happens without a person approving it once. Permission never lives in a prompt. Your agent proposes; you decide.
Why you can trust it
The agents here run on a real hash-chained ledger. Alter one line and every line below it fails verification. You don't have to take our word — recompute the whole chain in your own browser, right now.