> ## 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.

# Security Considerations

> Trust model, limitations, and compliance

# Security Considerations

## Trust Model

* **Verifier Trust:** Clients must trust the Verifier hardware (Intel TDX or AWS Nitro) and the Verifier operator's build process
* **Code Hash:** Organizations should verify image hashes through a trusted channel
* **Key Management:** Agent private keys must be stored securely; compromise allows impersonation

## What Spellguard Does NOT Provide

<Warning>
  **Message confidentiality at rest:** Archives contain encrypted payloads, but encryption keys must be managed separately.
</Warning>

* **Guarantee of message delivery** — AMP logs what happened, not what should have happened
* **Policy enforcement** — AMP audits; policy enforcement is a separate concern

## Standards Compliance

| Standard              | Usage                                      |
| --------------------- | ------------------------------------------ |
| RFC 9334 RATS         | Evidence building and verification pattern |
| A2A Protocol          | Agent discovery and interoperability       |
| Intel TDX / AWS Nitro | Hardware-backed Verifier attestation       |