Houdini

Problem Statement

Houdini serves as the definitive LLMS.txt storage for wallets, designed to give on chain AI agents easy access to structured identity and portfolio data. Instead of forcing AI models to guess context from raw transaction hashes, we utilize Octav.fi to retrieve accurate portfolio information and combine it with user verified inputs. This data is formatted into an llms.txt file the standard for AI readability and securely stored on 0g. By creating an Identity NFT that links to this file, Houdini ensures that any AI powered dApp or DAO can instantly access prepared, reliable knowledge about a user's on chain history.

Solution

Houdini is built on the 0G Network, combining a React frontend, Fastify backend, and Solidity smart contracts for Soulbound Identity NFTs. We deeply integrated the 0G Storage SDK to allow decentralized hosting of AI-readable llms.txt profiles. A major technical challenge was adapting the 0G SDK (designed for Node.js) to run entirely in the browser; we engineered a solution using Vite node polyfills to enable necessary crypto operations client-side. For data, we aggregate comprehensive user portfolios via the Octav API, processing the raw financial data into a semantic format. We also prioritized security UX by implementing ERC-7730 clear signing metadata, ensuring Ledger users see human-readable prompts like "Mint Personal Identity" instead of raw hex strings.