
VeriKraken AI is a trust-aware autonomous trading agent built for the AI Trading Agents hackathon. The project focuses on one core gap in AI trading systems: most bots are black-box and difficult to trust under real risk conditions. Our agent runs an autonomous decision loop using market checkpoints, applies strategy logic, and executes trades through a Kraken-style execution adapter. Every cycle is passed through explicit risk guardrails before execution. These controls include position-size limits, cooldown protection after trades, and drawdown-aware safety behavior to reduce over-trading and capital risk. To improve transparency and trust, VeriKraken AI generates ERC-8004-style validation artifacts for every checkpoint/trade action. Each artifact includes hashed intent and execution traces, creating an auditable proof trail. Runtime behavior is recorded in structured logs, making it easy for judges and users to verify what happened, when it happened, and why. Live MVP evidence already demonstrates: • successful ENTER_LONG execution • COOLDOWN ACTIVE guardrail enforcement • continuously generated validation artifacts • full runtime checkpoint and decision logs This project is designed as a practical foundation for trust-scored autonomous finance agents: safety-first, verifiable by default, and extensible toward on-chain identity/reputation workflows.
12 Apr 2026