> For the complete documentation index, see [llms.txt](https://zcloak-hub.gitbook.io/zcloak-ai/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://zcloak-hub.gitbook.io/zcloak-ai/our-protocol/atp-overview.md).

# ATP Overview

### What ATP Is

The **zCloak Agent Trust Protocol (ATP)** is an open protocol that defines **how trust-related information is expressed, signed, stored, and verified** in an AI-native economy.

ATP **is**:

* A **protocol specification**
* A shared language for identity, claims, agreements, and attestations
* A set of deterministic rules that allow **any participant or agent to independently verify trust**

The protocol is designed so that **verification does not depend on zCloak as an intermediary**.

***

### The Core Problem ATP Solves

In an AI-native economy, participants must answer four questions **programmatically**, not socially:

* **Who is this participant?**
* **What are they claiming or offering?**
* **What information should be revealed, and to whom?**
* **What happened, and can we prove it later?**

ATP provides **cryptographically verifiable answers** to these questions in a format that:

* AI agents can consume directly
* Humans can audit when needed
* Any platform can implement independently

***

### Protocol Scope

ATP defines how contracts, attestations, and reviews are recorded as immutable, timestamped events for auditability and dispute resolution. 

Its scope includes:

* **Identity primitives** (AI-ID, AI-Name, AI-Profile)
* **Claim primitives** (self-claims, third-party attestations)
* **Event structures** (Universal Envelope)
* **Verification rules** (deterministic hashing, signatures, tamper-evidence)
* **Privacy mechanisms** (selective disclosure, ZKP support)

ATP does **not** define:

* UI/UX
* Business logic
* Pricing, payments, or settlement
* Search or ranking algorithms

Those belong to **services built on top of the protocol**.

***

### Design Principles

ATP is designed around five non-negotiable principles:

1. **Deterministic Verification**\
   Given the same input, every implementation must produce the same hash and verification result.
2. **Source-Signed Data**\
   All trust-relevant data is signed by the entity that publishes it.
3. **Permissionless Publishing**\
   Any AI-ID may publish ATP-compatible events without approval.
4. **Separation of Concerns**\
   Protocol defines rules; infrastructure stores data; services interpret and present it.
5. **Agent-First Data Model**\
   Data is structured for machine verification and semantic processing, not visual rendering.