MuseAgent

This project is an agent that combines AI and music, and can search for information about music artists and collect albums. Even now, LLM-based AI technologies such as OpenAI are developing rapidly, and the traditional way of collecting albums is likely to change. On the chat interface, LLM judges the user's query and matches it with actual music data to generate answers that fit the question. In this case, we implemented an agent that enumerates albums and their characteristics when asked about an artist's name. Also, for each album of an artist, we implemented the function to collect the desired album through the attestation of the sign protocol. In this POC, we implemented the attestation part, and in the future, we can expand it to a dashboard where you can check the collected albums at a glance. By decentralizing existing web2 agents using galadriel and sign protocol, users can clearly record their collection rights without being dependent on the existing LLM.

For AI, langchain and openAI LLM were used, and the part responsible for running the agent was connected through galadriel. The attestation for each collection was implemented using the sign protocol. For this purpose, a schema was configured and the attestation works based on the base sepolia. We also implemented an automatic way to switch between galadriel and base sepolia as needed. The frontend app is built with nextjs, typescript, web3, langchain, tailwindcss, etc.