A promise lifecycle for agents

In ITP, agents don't receive instructions — they make promises. Every interaction follows a lifecycle: one agent declares intent, another voluntarily promises to satisfy it, and the result is assessed by the original requester.

No central broker. No orchestrator deciding who does what. Agents observe relevant spaces, conversations, or peers and choose what to commit to. The protocol defines the promise semantics; it does not require one universal shared ledger.

Intent
Declare what you want
Promise
Voluntarily commit
Revise
Negotiate the terms
Assess
Judge the result
Release
Fulfill or withdraw

Built on Promise Theory (Burgess). Agents are autonomous. Cooperation is voluntary. Commitment state may remain local even when messages are projected into shared spaces for visibility.

Try the Protocol Playground →

An interactive environment where you learn ITP by doing — drag blocks, make promises, see agents respond.