Sindri is a platform that makes compiling Zero-Knowledge Proof circuits in any framework and generating proofs with full GPU acceleration as easy and scalable as serverless platforms like AWS Lambda. The CLI tool offers an easy and intuitive interface for circuit development and deployment that will feel very familiar to anyone who has used Docker, Git, Heroku, or NPM.
For more information about the Sindri platform, please check out sindri.app. The best way to get started with the Sindri CLI is to begin with the Sindri CLI Quickstart and to then refer to the Sindri CLI Reference Docs for more detailed information about the CLI commands and options.
Sindri makes it easy to develop circuits in the framework of your choice, and to deploy them in a highly scalable and cost-effective way. You only pay for what you use, and you can scale up to thousands of GPUs in seconds. Our public circuit registry also makes it easy to share and reuse circuits, so you can build off of circuits that have already been audited and battle tested even if they're written in a different framework than the one you're using.
This section provides the essentials to get started with the CLI. The Sindri CLI Quickstart goes into more detail on each of these steps and is the recommended way to get started.
The Sindri CLI tool is available as an NPM package and can be installed with the following command:
npm install -g sindri@latest
To compile circuits on Sindri, you'll need to have an account and authenticate the CLI to use it. Visit sindri.app to find more details about account creation.
sindri login
The Sindri CLI provides a project scaffolding tool to help you get started with circuit development.
The sindri init
command will initialize a new circuit project for you with everything you need to get started.
sindri init my-circuit
You can run sindri lint
from within your circuit project directory to perform local checks to ensure your circuit is valid and ready for compilation.
This will not actually compile your circuit, but will perform basic checks to uncover issues that would prevent your circuit from compiling successfully.
sindri lint
To compile your circuit on the Sindri platform, you can use the sindri deploy
command to upload and compile it.
If there are compilation errors, they will be reported, and the command will exit with a non-zero exit code.
sindri deploy
Most of the project commands should be relatively easy to run natively on any system with node and Yarn v1 installed, but a Docker Compose configuration is provided as a convenience.
In addition to ensuring that the proper dependencies are available, the container is also configured to install the package globally as a symbolic link with npm link
and automatically rebuild it whenever files change.
This means that the sindri
command inside the container should always reflect your local code changes.
You can bring the container up in the usual way.
docker compose up
This will install the latest packages and start the build process in watch mode so that it rebuilds whenever local files change.
To drop into a shell in the running container, you can run the following.
docker compose exec sindri-js bash
From here, you can use the sindri
command or run any of the Yarn invocations for linting, formatting, etc.
Inside of the Docker container, you can run
sindri -d login -u http://host.docker.internal login
to authenticate against a Sindri development backend running locally on port 80.
Both your credentials and the local base URL for the server will be stored in sindri.conf.json
and used in subsequent requests.
To install the project dependencies:
yarn install
To run the build process in development mode and watch for file changes:
yarn build:watch
This command will run automatically in the container if you're using Docker Compose.
To perform a production build with minified outputs:
yarn build
The test suite can be run with:
yarn test
This will first build the project and then run the tests against the build outputs using pre-recorded fixtures for network requests.
To skip the build step, you can use yarn test:fast
.
In development, you can use
yarn test:watch
to rebuild the project and rerun the tests whenever there are code changes. Additionally, this command will allow network requests without fixtures to go through, so this is useful when writing new tests.
This command will build the project and record new fixtures for all network requests.
yarn test:record
You will need to have your environment authenticated with an API key for the "Sindri CLI/SDK Testing Team" organization on Sindri in order to make the necessary network requests for recording fixtures.
The credentials stored by sindri login
will be used automatically for the tests.
This will fetch the latest OpenAPI schema for the Sindri API and autogenerate an updated API client in src/lib/api/
.
yarn generate-api
To develop against unreleased API features, you can use these variants to target a local development server:
# If you're not using Docker:
yarn generate-api:dev
# Or...
# If you are using Docker:
yarn denerate-api:docker
The Sindri Manifest JSON Schema is stored in sindri-manifest.json
and needs to be manually updated and committed when the schema changes.
The file can be updated by running:
yarn download-sindri-manifest-schema
To develop against an unreleased version of the schema, you can use these variants to target a local development server:
# If you're not using Docker:
yarn download-sindri-manifest-schema:dev
# Or...
# If you are using Docker:
yarn download-sindri-manifest-schema:docker
To lint the project with Eslint and Prettier:
yarn lint
This check will already run on CI.
To reformat the project using Prettier:
yarn format
To check the TypeScript types:
yarn type-check
This check will already run on CI.