VEX Core

Core types for the VEX protocol — adversarial, temporal, cryptographically-verified AI agents.

Key Types

• Agent — Fractal agent with evolutionary capabilities
• ContextPacket — Time-aware, hashable context unit
• MerkleTree — Cryptographic verification of context hierarchies
• Genome — Evolutionary traits that map to LLM parameters

Quick Start

use vex_core::{Agent, AgentConfig};

// Create an agent
let agent = Agent::new(AgentConfig {
    name: "Researcher".to_string(),
    role: "You are a helpful research assistant".to_string(),
    max_depth: 3,
    spawn_shadow: true,
});

// Spawn a child agent
let child = agent.spawn_child(AgentConfig {
    name: "Specialist".to_string(),
    role: "You analyze data".to_string(),
    max_depth: 2,
    spawn_shadow: false,
});

Merkle Verification

use vex_core::{MerkleTree, Hash};

// Build a Merkle tree from context hashes
let leaves = vec![
    ("ctx1".to_string(), Hash::digest(b"context 1")),
    ("ctx2".to_string(), Hash::digest(b"context 2")),
];
let tree = MerkleTree::from_leaves(leaves);

// Verify integrity
assert!(tree.root_hash().is_some());

vex-core

Core types and primitives for the VEX (Verified Evolutionary Xenogenesis) Protocol.

Features

• Agent - Autonomous AI agent with cryptographic identity
• ContextPacket - Immutable, versioned context for agent memory
• MerkleNode - Cryptographic verification of agent state history
• Evolution - Trait-based agent evolution and improvement tracking

Installation

[dependencies]
vex-core = "0.1"

Quick Start

use vex_core::{Agent, ContextPacket};

#[tokio::main]
async fn main() {
    let agent = Agent::new("my-agent");
    let context = ContextPacket::new("Initial context");
    // ... use agent
}

License

Apache-2.0 License - see LICENSE for details.