> ## Documentation Index
> Fetch the complete documentation index at: https://docs.spellguard.ai/llms.txt
> Use this file to discover all available pages before exploring further.

# Architecture

> How Spellguard provides unified governance for your AI agents

# Architecture Overview

Spellguard provides a unified control plane for managing AI agents across your organization. Register agents from any vendor, enforce consistent policies, and maintain complete audit trails — all from a single dashboard.

For agent-to-agent communication, Spellguard acts as a neutral verification layer. When two agents want to communicate, both verify that the Spellguard Verifier is running the expected code, then register their identities with cryptographic evidence. Spellguard combines **cTLS** (for mutual verification and secure channels) with **AMP** (for auditable message logging) so that every interaction is both private and provable.

```mermaid theme={null}
flowchart LR
    A[AGENT A] -->|"1. Verify & Register"| SG

    subgraph SG ["SPELLGUARD VERIFIER"]
        direction TB
        cTLS["cTLS: Attestation & Keys"]
        cTLS --> AMP["AMP: Commitments & Logging"]
    end

    SG -->|"2. Secure Channel"| B[AGENT B]
```

## How It Works

1. **Agents verify the Verifier** before connecting by checking its attestation document
2. **Agents register** with signed evidence proving their identity
3. **Agents communicate** through encrypted channels with full audit trails

<Info>
  The Verifier never sees plaintext message content. All payloads are encrypted client-side before transmission.
</Info>