Core Architectural Principles

We first introduce the core principle/phrases in this page.

Architecture Overview

+--------------------------+
|     Solana / SVM Chain   |
+--------------------------+
            ^
            | Attestation, Proofs
            v
+--------------------------+
|    Nada Verification     |  <-- On-chain program verifying TEE proofs
+--------------------------+
            ^
            | Execution Receipts
            v
+--------------------------+
|      TEE Execution       |  <-- Agents run inside SGX/Nitro/SEV
|  Nada Agent Runtime      |
+--------------------------+
            ^
            | Input, Prompts, Tasks
            v
+--------------------------+
|        User / dApp       |
+--------------------------+

1. Confidentiality

Inputs, execution state, and outputs of agents remain private inside TEEs.

2. Verifiability

SVM programs can verify:

• TEE vendor attestation signatures

• enclave measurement (hash of binary)

• enclave-bound public key

• execution provenance (context & attention proofs)

• output signatures

3. Deterministic Verification, Non-Deterministic Execution

The TEE produces nondeterministic computation; the blockchain verifies a deterministic proof.

4. Compatibility

No changes required to:

• Solana runtime

• Solana clients

• Solana development tools

• existing SVM programs

Nada functions as a modular confidentiality layer.