DeFAI Agent Marketplace. Browse, fork, copy & earn royalties from intelligent iNFT agents.
GhostFi is a DeFi Strategy Agent Marketplace that lets anyone truly own autonomous AI agents as iNFTs powered by 0G’s ERC-7857 standard.
Users can browse a live Arena of AI agents ranked by yield and TVL, fork top strategies, or create their own in one click. Each agent runs 24/7: scanning pools, optimizing yields, restaking, or lending autonomously using multiple AI models and verifiable inference on 0G Chain.
What makes GhostFi special is real ownership: when you mint an agent as an iNFT, you own its encrypted intelligence, strategy logic, and performance history. You can trade the iNFT, earn royalties when others copy your agent, or transfer the complete AI agent to a new owner.
Every decision is attested on-chain as Proof of Intelligence, while strategy logic stays private until executed. GhostFi turns passive DeFi users into creators and owners of valuable AI-powered yield strategies.
We built GhostFi as a full-stack hackathon prototype using a React + Vite frontend paired with a Python Flask backend.
The frontend delivers a fast, responsive neo-brutalist UI with smooth animations. It includes the Landing page, Marketplace, Arena leaderboard, one-click Agent Creation flow, Pool Details, My Agents, and rich iNFT views. Users can easily browse, fork, configure, and mint agents.
The backend exposes clean REST endpoints for scanning and deploying agents. It handles data orchestration, AI strategy selection, and execution logic. Core Technologies & Integration:
One notable hack was mixing live protocol APIs with local mock fallbacks and simulated deployment results. This allowed us to create a reliable, fully functional end-to-end demo that never breaks, even when external services are rate-limited or unstable during the hackathon.
The entire flow: from scanning opportunities with Uniswap data → Groq AI decision → 0G verification → Hedera Agent Kit orchestration → safe execution — is designed to feel instant and magical while remaining modular for future development.

