Decentralized Autonomous Civilization

Introduction

The Decentralized Autonomous Civilization (DAC) is a Solana-based network that enables verifiable, decentralized execution of AI agent tasks with granular payment management. The protocol supports both public nodes (standard execution) and confidential nodes (TEE-enabled for private data processing), implementing a multi-validator consensus system to ensure trust and correctness.

Key Features:

• Dual Node Architecture: Public nodes for standard tasks, confidential nodes (Intel SGX TEE) for private data processing
• Multi-Validator Consensus: Agents, nodes, and tasks require multiple validations before approval/processing
• Share-Based Payment System: Proportional cost distribution with automatic refunds and granular contribution tracking
• Goal-Oriented Execution: Define objectives with iteration limits, treasury funding, and agent assignments
• Cryptographic Integrity: SHA256 chain proofs verify complete execution history
• Decentralized Validation: Any active node (public or confidential) can validate task execution results

Core Components

Smart Contract (DAC)

The program running on Solana that manages all state objects and operations.

Program Address (Devnet): BaY9vp3RXAQugzAoBojkBEZs9fJKS4dNManN7vwDZSFh

Node Types

• Public Nodes: Standard nodes that can execute public tasks and validate any tasks/nodes
• Confidential Nodes: Run in Intel SGX TEE (Trusted Execution Environment), can execute both public and confidential tasks, and validate any tasks/nodes
• Key Distinction: Only confidential nodes can CLAIM/EXECUTE confidential tasks (TEE protection for private data). However, any active node (public or confidential) can VALIDATE task execution results and other public nodes.

Data Storage

• IPFS: All task data and configurations files stored off-chain
• On-Chain: Only IPFS CIDs and cryptographic proofs stored

Key Workflows

1. Network Initialization: Authority initializes network with approved code measurements and pre-allocates goals and tasks
2. Node Registration:
   • Public nodes register and must pass benchmark validation by multiple validators
   • Confidential nodes register and prove TEE hardware (self-approved via TEE attestation)
3. Agent Setup: Create agents with validated configurations stored on IPFS (requires multi-validator consensus)
4. Goal Creation: Define objectives with treasury funding and iteration limits (can be public or confidential)
5. Task Execution:
   • Nodes claim tasks (confidential tasks only claimable by confidential nodes)
   • Nodes execute tasks with LLM
   • Any active node validates execution results (multi-validator consensus required)
   • Payments triggered when validation threshold reached

Security Features

• TEE Attestation: Confidential nodes prove genuine Intel SGX hardware
• Code Measurement Whitelist: Only pre-approved code can run in TEE
• Multi-Validator Consensus: Agents, nodes, and tasks require multiple validations before approval/processing
• TEE Signatures: Confidential task validations cryptographically signed (only confidential nodes can provide)
• Chain Proofs: SHA256 hashes verify data integrity across all executions
• Access Control: Confidential goals can only be claimed/executed by confidential nodes (TEE protection)

Building and Testing

Build

Build the Solana program and generate clients:

anchor run build

This command:

• Compiles the Solana program
• Generates TypeScript and Rust clients in clients/ directory
• Creates the program IDL

Test

Run the test suite:

anchor run test

Or use the verbose flag directly:

anchor test run --verbose

Documentation

For detailed architecture, requirements, and specifications, see:

• Design Document - Complete architecture and design specifications
• User Stories - Detailed user stories and technical implementation