Summary

HammerGPT/Hyper-Alpha-Arena provides an open-source, production-ready platform for autonomous cryptocurrency trading using Large Language Models (LLMs). It targets engineers and power users seeking to deploy AI-driven trading strategies, offering both risk-free paper trading simulations and live perpetual contract execution on the Hyperliquid Decentralized Exchange (DEX). The platform aims to democratize advanced AI trading by integrating LLMs directly into the trading workflow with robust risk management features.

How It Works

The platform leverages LLMs (e.g., GPT-5, Claude, Deepseek) to autonomously generate and execute trading strategies. It ingests real-time market data via the ccxt library and processes it through customizable AI prompts. Users can deploy strategies in a simulated paper trading environment or directly on the Hyperliquid DEX via perpetual contracts, supporting leverage up to 50x. The architecture emphasizes secure private key management and environment isolation between testnet and mainnet.

Quick Start & Requirements

• Primary Install: Clone the repository and run docker compose up -d (or docker-compose up -d). The application will be accessible at http://localhost:8802.
• Prerequisites: Docker Desktop (Windows, macOS, Linux).
• Links:
  • Docker Desktop: https://www.docker.com/products/docker-desktop/
  • Hyperliquid Docs: https://hyperliquid.gitbook.io/

Highlighted Details

• Multi-Model LLM Support: Integrates with any OpenAI API-compatible language model, with Deepseek recommended for cost-effectiveness.
• Hyperliquid DEX Integration: Full support for perpetual contract trading on both testnet and mainnet, including market/limit orders and 1-50x leverage.
• Advanced Risk Management: Features configurable maximum leverage, margin usage alerts, automatic trading pauses at 80% margin, and liquidation price monitoring.
• Prompt Template System: Offers customizable AI trading prompts with a visual editor and account-specific binding for tailored strategies.

Maintenance & Community

The project is actively developed, with Phase 2 (Hyperliquid Integration) completed and Phase 3 (Enhancement & Optimization) in progress. Key contributors are acknowledged from the original open-alpha-arena project and inspiration sources like nof1.ai. Community support is available via a knowledge planet (link provided in README) and direct contact on Twitter/X (@GptHammer3309). The official website is https://www.heliki.com/.

Licensing & Compatibility

This project is licensed under the Apache License 2.0. It requires attribution to the Heliki AI Community in documentation or product notices. The license is generally permissive for commercial use, provided attribution and license terms are met.

Limitations & Caveats

Following the v0.5.0 update, historical trading records lacking wallet addresses will not appear in wallet-specific UI views; manual data migration is required for advanced users. The platform currently recommends using only one AI model per Hyperliquid wallet address for accurate performance tracking, with multi-model support for a single wallet planned. Trading is primarily focused on BTC pairs, with support for additional pairs slated for future development.