The problem Reputation MicroRollup solves

The entirety of lending and borrowing in DeFi runs on overcollaterized loans. This not only is capital-inefficient but also prevents smaller players from being able to take enough credit. A web2 approach to this problem is having credit scores that considers the entire borrowing and lending history of a person to offer calculated terms on the loans. This can be done on chain too considering protocols like Aave and Compound have existed for a long time and people have built up a large, verifiable, public credit history. Some other protocols have tried to implement this but we have made significant improvements over the previous approaches.

1. The credit history data is fetched through multiple sources across chains ensuring that a wide variety of information is available to accurately calculate the credit score. This allows for chains that do not have a dexes, or lending protocols to import creditworthiness without a prior DeFi ecosystem.
2. The score calculation and risk engine is run offchain with the results being posted on chain through Stackr. This makes it possible to run complex mathematical models or even ML and AI models on the data which would be expensive on chain. All the data is verifiable too on chain.
3. The credit score that everyone is assigned has an associated vouching power too. With this, people with a better credit score can vouch for someone with a lesser credit score to help them secure better loans. This creates a market for reputation-as-a-service (RaaS) where people can stake their reputation on someone else to earn rewards if they are able to pay off their loans.
4. It is also composable meaning that other people can easily add newer sources of fetching credit history and a whole host on new protocols can now be built on this other developers to use this novel method to build higher-level protocols to offer undercollaterized loans or better APY rates. There are a lot of other usecases that developers can now build using this as base layer.

Challenges we ran into

We are using Stackr and 1inch token plugins in our project which both had their fair share of challenges. Stackr SDK is quite new and there were bound to be issues (for example using a new paradigm of programming, some signature issues etc.) but we managed to solve all of them thanks to the constant support of Stackr devs and mentors.

1inch token plugins too used an older version of OpenZeppelin's contracts which we found out about quite late. We had some breaking changes but resolved them with correct versioning and looking at the other plugins that had been built.