The Coinbase Prime Liquidator sample application continuously monitors a Coinbase Prime portfolio for crypto assets in hot/trading wallets and places sell orders or converts to USD/fiat.
Sell orders deduct holds based on instrument, so if new assets are added while others are being liquidated, the application continues to create new orders if there are enough assets to sell. Additionally, if for some reason an order continuously fails to execute, there is an ID generated (client_order_id) from the sell attributes that is used to reduce the amount of spam/failing orders sent to Prime. Prime treats the client_order_id as idempotent for open orders and the ID is cached in-process too.
The Coinbase Prime Liquidator sample application is free and open source and released under the Apache License, Version 2.0.
The application and code are only available for demonstration purposes.
Use of this sample application may cause a negative financial impact
When the application is running, it continuously monitors and converts crypto assets to USD. If the application is accidentally left running or mistakenly pointed at an unintended portfolio, all of the assets in hot/trading wallets will be quickly liquidated.
Sell orders are created with a one hour TWAP with the lowest tolerance (limit price) set at 10% below the current price of the asset price on the Coinbase Exchange.
If the sample application is used to liquidate large positions, there is price action risk that may result in trades executing up to 10% lower than the latest price check.
If this application is deployed using the sample AWS CloudFormation template, there will be new charges to your AWS account. For high throughput Prime portfolios, these charges may be significant. As always, continiously review your AWS bill to understand more.
The Coinbase Prime Liquidator has a sample AWS CloudFormation (CFN) template that can be deployed to run the application in an Amazon Elastic Container Service (Amazon ECS) cluster. This template creates all of the required resources and can be customized to suit the deployers needs. To deploy the CFN stack, initialize your AWS credentials and then run:
make create-aws-stack ENV_NAME=dev PROFILE=default REGION=us-east-1Customize the values of the ENV_NAME, PROFILE, and REGION to match the needs of your environment. The PROFILE argument is the name of the AWS CLI profile configured and the REGION argument is the AWS Region you would like to deploy to.
Once the CFN stack is deployed, configure your credentials in AWS Secrets Manager. The name of the empty secret uses the following format:
prime-liquidator-ENV_NAME-prime-api-credentials
The ENV_NAME will the same as what was passed to the create-aws-stack command (e.g., dev). The credentials use the following format:
{
"accessKey": "",
"passphrase": "",
"signingKey": "",
"portfolioId": "",
"svcAccountId": ""
}
Prime API credentials can be created in the Prime web application, once an account is opened.
The inital stack deploys the public.ecr.aws/nginx/nginx:stable-perl-arm64v8 container as a placeholder. The ECS task definition/service requires a container and at this point, the ECR repository has not been created. Once the CFN stack is deployed, build the container image and deploy to the Amazon ECR (ECR) repository. To do this, execute:
make build-image ENV_NAME=dev PROFILE=default REGION=us-east-1 BUILD_ID=1Again, customize the values of the ENV_NAME, PROFILE, BUILD_ID, and REGION to match the needs of your environment. The BUILD_ID can be set to a value that matches your container tagging practices.
After the build is deployed, update the DockerImageUri attribute in the aws-dev.json file to the URI of your deployed build. The newly created ECR respository uses the following format:
AWS_ACCOUNT_ID.dkr.ecr.AWS_REGION.amazonaws.com/prime-liquidator-ENV_NAME:BUILD_ID
Note: If your environment is not named, dev, you will need to create a new environment configuration file. The environment name is defined by the ENV_NAME CLI argument. The format for the environment configuration file name is:
infra/aws-ENV_NAME.json
The CFN stack is configured to run ARM64 containers, so you must build on an ARM64 compatible computer.
Once the DockerImageUri attribute is updated, run the following command to update the CFN stack and run the Coinbase Prime Liquidator:
make update-aws-stack ENV_NAME=dev PROFILE=default REGION=us-east-1This command deploys the container image specified in the DockerImageUri and starts listening for new Prime activities.
To build the sample application, ensure that Go 1.21+ is installed and then run:
go build cmd/server/main.goTo build the Docker container, login to the Amazon ECR Public Gallery:
aws ecr-public get-login-password --region us-east-1 | docker login --username AWS --password-stdin public.ecr.awsRun the docker build:
docker build .