Skip to content

Latest commit

 

History

History
290 lines (204 loc) · 10.2 KB

README.md

File metadata and controls

290 lines (204 loc) · 10.2 KB
Raw

Aave Starknet Bridge

Tests Check License: MIT

⚠️ This codebase is still in an experimental phase, has not been audited, might contain bugs and should not be used in production.

Table of contents

Introduction

For Aave, one of the main current and future goals is growth of liquidity and user base. As seen on side-chains with low transaction cost like Polygon or Avalanche, there is high demand to use the Aave protocol with small amounts to earn high yield. That's why we brought to you an initial phase of the Aave <> Starknet integration allowing deposit/withdrawal on Aave Ethereum by exclusively transacting on Starknet.

The bridge allows users to deposit and withdraw their aTokens, and only aTokens, on StarkNet and get static_a_tokens - wrappers converting balance-increasing aTokens into exchange-rate-increasing static_a_tokens. We assume that L1 tokens approved by the bridge are pre-validated tokens, and that they are not deflationary.

The bridge is also shaped for liquidity providers who are able to assume the Ethereum gas cost of deposits and withdrawals as they transact large enough amounts. They will deposit on Aave Ethereum, bridge the static_a_tokens to Starknet and make them available for users there to buy and hold, accruing this way yield from L1.

Architecture

aave_bridge

Contracts

Overview

L1

  • Bridge - handles rewards update, deposit & withdrawal of static_a_tokens, their corresponding aTokens and their underlying assets
  • Proxy - A proxy implementation

L2

  • static_a_token - exchange-rate-increasing wrapper of aTokens on Starknet
  • incentivized_erc20 - tracks users' claimable rewards and current reward index for each static_a_token
  • rewAAVE - ERC20 representing the rewards on L2
  • bridge - bridge responsible for:
    • minting and burning static_a_tokens on message from L1
    • bridging rewAAVE tokens back to L1
    • updating rewards_index for each static_a_token on message from L1
  • proxy - generic implementation of a proxy in cairo

More about static_a_token on L2

Natively, Aave tokens grow in balance, not in value. To be able to create this kind of model, it is important to wrap them before bridging, converting them in a token that grows in value, not in balance.

static_a_tokens are an implementation of the wrapped aTokens that will continuously increase in value on Starknet because they are backed by the increasing aTokens amounts locked in the bridge contract on Ethereum. static_a_tokens can then be bridged back to aTokens.

Proxies

All calls made to the following contracts will be handled by a proxy who delegates the calls to the available implementation of these contracts.

  • bridge on L2
  • static_a_tokens on L2
  • Bridge on L1
  • rewAAVE token on L2

Governance

  • static_a_tokens are controlled by L2 bridge.
  • rewAAVE token is owned by L2 bridge.

How does it work?

Bridging aTokens from L1 to L2

Approve bridge tokens

L1 aTokens are approved on the bridge at initiliaze where _approveBridgeTokens is called internally to approve the provided array of aTokens in an array along with their corresponding static_a_tokens on L2. ⚠️ Gas limit concerns should apply here!

Transfer from L1 to L2

Users can either bridge their aToken (let's say aDai) or deposit the underlying asset (i.e Dai). Users will have to approve the bridge to spend the underlying asset tokens or aTokens, depending on the provided value for fromUnderlyingAsset argument when depositing.

Calling deposit allows users deposit aTokens or their underlying asset:

If depositing underlying asset:

  • The asset token will be transferred from the user account to the L1 bridge.
  • The bridge will then deposit the asset token in the aToken.
  • A message will be sent to the L2 bridge with the amount of static_a_token to be transferred, the L1 token address, the recipient address, the block number and the rewards index.
  • The token bridge on L2 will then be minting the corresponding static_a_token of the L1 token to the user.

If depositing aTokens:

  • The aTokens will be transferred from the user account to the L1 bridge.
  • A message will be sent to the L2 bridge with the amount to be transferred, the L1 token address and the recipient address as parameters.
  • The token bridge on L2 will then be minting the corresponding static_a_token of the L1 token to the user.

Transfer L2 to L1

To bridge their aTokens back to L1, users need to initiate a withdrawal on the L2 token bridge.

Calling initiate_withdraw will result in the following:

  • The amount to withdraw will be burned by the bridge
  • A message will be sent to L1 with the L1 token address, the L1 recipient, the L2 rewards index and the amount
  • The L1 bridge will then transfer the aTokens to the L1 recipient
  • The L1 bridge also checks for any difference in the L1/L2 rewards index and transfers any unclaimed rewards to the L1 user

Synchronisation of rewards on L1 and L2

Starknet users will continue to enjoy the same rewards as on L1 after bridging their assets. To achieve that we continuously update the rewards_index of all static_a_tokens to match the value of their respective aTokens on L1, by tracking the reward index on departure of the static_a_token and sending the rewards accrued during the bridging process to the recipients address.

Claiming rewards on L2

To claim rewards users need to call claim_rewards on static_a_token contract which calls the bridge in return to mint the due rewAAVE tokens to the user.

Bridging rewards from L2 to L1

Calling bridge_rewards on L2 token bridge results in:

  • The bridged amount of rewAAVE tokens will be burned.
  • The L1 bridge receives the bridging message and claims the rewards amount to self by calling claimRewards on the IncentivesController contract.
  • The rewards are then transferred to the L1 recipient.

Installation

Environment

Install Node 16

Our codebase relies on Node 16. To install it, you can first install nvm and then run the following commands:

nvm install 16
nvm use

Install Python 3.7.12

Our codebase relies on Python 3.7.12. To install it, you can first install pyenv and then run the following commands:

pyenv install 3.7.12
pyenv local 3.7.12

Install GMP (needed for Cairo)

Before installing Cairo you need to install GMP. Run one of the following command depending on your OS.

sudo apt install -y libgmp3-dev # linux
brew install gmp # mac

Install Node dependencies

Let's install all our project dependencies:

yarn install

Install Python dependencies

Let’s create a virtual environment to isolate your project’s requirements from your global Python environment.

python -m venv .venv
source .venv/bin/activate

Install poetry for dependencies management

python -m pip install --upgrade pip
pip install poetry
poetry install

Build Cairo files

Solidity files are automatically compiled before running the tests, but Cairo files are not. To compile them, run:

yarn compile:l2

Start testnets

We recommend to run L1 and L2 testnets in different terminals.

Start L2 testnet

In a terminal where venv is activated, run:

yarn testnet:l2

Start L1 testnet

Create a .env file from the sample (cp .env.sample .env), and fill a value for the variable ALCHEMY KEY - you can get one here. Then, load all the environment variables.

source .env

And start L1 testnet in the same terminal by running:

yarn testnet:l1

Run tests

The project is tested using hardhat, the starknet hardhat plugin and starknet-devnet. We created a Docker Compose file to run tests easily: we start L1 and L2 test networks in two separate containers and run the tests from a third one. To run all tests, simply run the following commands:

docker compose up --build
docker exec -ti $(docker ps -f name=test-runner -q) bash
yarn test

Deployment

First make sure to set the aTokens addresses to be approved on the bridge as well as the metadata related to the staticATokens to be deployed on l2 in ./scripts/allowlistedTokens.ts.

yarn deploy-bridge:testnet #deploys bridge on l1 & l2 testnets

Contributors