Harolds agent v2

**Harold** is a fully autonomous AI trading agent that solves the trust problem in algorithmic trading. While most bots operate as opaque black boxes — executing trades with zero accountability — Harold cryptographically signs every trade decision using EIP-712, anchors his identity and reputation on the Sepolia testnet via ERC-8004, and gates all executions through a whitelisted Risk Router contract. Every move is auditable, every intent is verifiable, and no trade can go rogue. Under the hood, Harold ingests live Kraken price feeds and Prism API news sentiment every 5 minutes, runs it through a Groq-powered LLaMA 3.3 70B brain for sub-second inference, and produces structured JSON trade decisions with built-in risk controls — 1% take profit, 1% stop loss, SMA alignment gates, and volume spike detection. The chain enforces his rules so he literally cannot break them. Harold isn't just a trading bot. He's a proof-of-concept for what trustless autonomous agents look like — where you don't have to trust the developer, the model, or the team. You verify the contract, check the signature, and cross-reference the tx hash on Etherscan. That's it. **Built for the lablab.ai AI Trading Hackathon. Competing on both the Kraken and ERC-8004 tracks.**