The project mainly consists of three modules.

The Finder module scans potential staking contracts and applies basic filtering. The filtering rules are:

• If the contract can act as a recipient of ERC-20 tokens and has a positive balance, it will be verified once;

• after a certain number of blocks, if it is verified again, it will be classified as verified contracts.

The Risk module analyzes risks of verified projects, including(unfinished):

• Yield risk: excessively high yields or overly short reward duration;

• TVL concentration risk: TVL is overly concentrated;

• Privilege risk: the owner can withdraw funds, the contract can be paused or frozen, and rewards can be arbitrarily modified;

• EOA risk;

• Unlimited reward token minting;

• Short-term extreme TVL volatility;

• No history or very new contracts.

The Sink module is mainly responsible for interacting with ClickHouse.

The project is built in Rust, primarily using the Alloy library to interact with on-chain data, and the ClickHouse library to interact with ClickHouse.

ChatGPT performed exceptionally well throughout this process—not only in helping complete the code, but also by providing strong design suggestions thanks to its extensive blockchain domain knowledge.

This project only supports staking projects that accept ERC-20 contracts as deposit tokens. This constraint was intentionally chosen to keep the MVP as simple as possible; however, it overlooked the fact that such projects are relatively rare. As a result, it was difficult to find any project that met the criteria, and a significant amount of time was spent debugging the first part. Consequently, the second part—the risk scanning—was not completed on schedule.