simplescan.ai
Website Link
GitHub Link
Project Motivation
Etherscan provides the capability to trace any address or transaction on the Ethereum blockchain. However, there is a notable absence of a systematic method to classify smart contracts or understand their function within the ecosystem. Furthermore, Etherscan does not allow for much accessibility to those without extensive technical knowledge. Enter simplescan.ai!
What Is It? ****
simplescan.ai is website that allows users to understand the purpose and function of smart contracts without actually reading the code. Below you can see the homepage of our website:
Users can simply input an address into the search bar above, and they are then taken to a results page that contains:
- A summary of the smart contract, along with a link to interact with the contract on Etherscan
- The category and subcategory that the contract falls into
- Descriptions of the contract’s read and write functions, along with links to interact with these functions
(see below!)
How We Built It
The project’s backend consists of three primary features and utilizes the following backend capabilities:
- Classification — The program classifies a given smart contract into category and subcategory in the Analysis module, and utilizes GPT-4.
- Summary — The purpose of the smart contract is broadly summarized in the Analysis module, and utilizes Google Gemini.
- Functions — The read and write functions are fetched and each of the function’s purposes are stated in the Scanner Module utilizing Etherscan API and in the Analysis module using GPT-4.
The project’s system architecture and flow of interactions can be found in the below diagram:
How to Use
- Go to https://simplescanai.onrender.com/
- Enter desired contract address in search bar (example address: 0xBB9bc244D798123fDe783fCc1C72d3Bb8C189413)
- Give the page a moment to load
- Read the displayed results!
Future Goals
- Provide information for addresses that are not smart contracts, such as wallets or transactions
- Decrease the latency of our application by designing our own classification mode, rather than using Chat GPT-4