Digital gold services depend on transaction histories because every purchase or sale changes a user's rupiah and gold balances. A centralized database supports efficient operations, but it does not by itself provide users with an independently verifiable record outside the operator's system. This study developed a web-based prototype that uses Ethereum Sepolia as a supplementary evidence layer for approved digital gold transactions. The study followed an applied system-development approach and the Waterfall model, supported by a literature review and observation of common digital gold workflows. The prototype separates end-user and administrator roles. Buy and sell requests remain pending until administrative review; rejected requests do not alter balances, whereas approved requests are submitted through MetaMask to a Solidity smart contract. The resulting transaction hash and block number are retained with the operational history in MySQL. The application was implemented with React, TypeScript, Tailwind CSS, Node.js, and Express.js. Black-box testing covered 24 application scenarios and nine blockchain-integration scenarios. All 33 scenarios passed, including insufficient-balance handling, role separation, approval and rejection states, purchase and sale recording, explorer verification, and consistency between the application ledger and Sepolia evidence. The results demonstrate the feasibility of using a public testnet as an additional transaction-evidence layer rather than as a replacement for the operational database. However, the prototype uses simulated balances and prices and does not establish the existence of physical gold or production-level security, scalability, and economic viability.
Copyrights © 2026