FindETH
Brands/creators launch internet treasure hunts with on-chain prizes and collectibles.
Demo
About
FindETH is a Web3 treasure-hunt platform where businesses and creators turn their websites, products, and campaigns into interactive onchain discovery experiences. Creators/brands use AI to generate branded hunts with clues, locations, difficulty, prizes, and answer hashes. The AI workflow queries live The Graph analytics before generation, using historical completion rates, prize sizes, clue counts, and category performance to recommend better hunt designs. Hunts are published through smart contracts on Sepolia, with clue answers protected as hashes and winners receiving verifiable NFT trophies. The platform combines wallet-based participation, Chainlink-powered mystery rewards, and backend metadata storage without exposing plaintext answers. FindETH also uses ENSv2 to create hierarchical namespaces for creators, hunts, and AI agents. Each hunt can receive an ENSv2 subname containing metadata records, while Permissioned Resolvers and Enhanced Access Control allow creators or delegated AI agents to update only the records they are authorized to manage.
How it's made
FindETH is built as a full-stack Web3 application. The frontend uses React, TypeScript, Vite, Tailwind CSS, Wagmi, Viem, and RainbowKit for the interface, wallet connection, contract interaction, ENS resolution, and Sepolia network support. Foundry and Solidity power the TreasureHunt and TreasureNFT contracts, including hashed clues, creator verification, prize handling, and Chainlink VRF-based mystery rewards. The backend is a FastAPI service using SQLAlchemy and Alembic with PostgreSQL/Supabase support. Gemini generates hunt drafts from real business content, while The Graph provides live onchain analytics before each generation so prize sizes, difficulty, and clue counts are informed by previous hunt performance. ENSv2 is integrated at the namespace level: creators receive hierarchical subnames for hunts and AI agents, with metadata stored in resolver text records. Permissioned Resolver and Enhanced Access Control enable delegated, record-specific updates. A notable security decision is that plaintext answers are never persisted in the backend. Only normalized answer hashes are stored onchain, while answers exist temporarily in the creator’s browser before publication.