Ajar Protocol
Concepts

Concepts

Plain-language pages that explain the problems Ajar handles and how the protocol resolves them.

This section explains Ajar from the problem outward. The protocol spec is the source of truth for exact objects and requirements. These pages explain why those objects exist, what used to go wrong without them, and how Ajar changes the flow.

Read these when you want the reasoning before the fields.

PageWhat it explains
Problem statementWhy scraping, forms, browser sessions, and weak proof are not enough
Owner contractHow a signed manifest lets the website owner say what agents may read or do
Discovery and verificationWhy indexes help discovery but the origin manifest remains the authority
Public, private, and authenticated accessWhy public reads can be simple, while account data needs delegated authority
Delegated authorityHow mandates replace broad session sharing with scoped, revocable permission
Model safetyWhy the model proposes while deterministic code enforces authority
Action safetyHow risk classes, SIMULATE, Offer, and Commit reduce action mistakes
Receipts and accountabilityHow signed receipts leave evidence after consequential actions
Metering and settlementHow Ajar declares price and settlement without becoming a payment rail
End-to-end flowHow the pieces fit together from owner setup to receipt

The short version: Ajar does not try to make agents better at guessing websites. It gives websites a way to publish a contract that agents can verify before they read, act, or pay.