Skip to main content
info

Please note that zkApp programmability is not yet available on Mina Mainnet, but zkApps can now be deployed to Berkeley Testnet.

How to Write a zkApp

A zkApp consists of a smart contract and a UI to interact with it. First, we’ll install the Mina zkApp CLI and write a smart contract.

Write a smart contract

Your smart contract will be written using Mina zkApp CLI.

Mina zkApp CLI makes it easy to follow recommended best practices by providing project scaffolding including dependencies such as SnarkyJS, a test framework (Jest), code auto-formatting (Prettier), linting (ES Lint), & more.

Install Mina zkApp CLI

npm install -g zkapp-cli

Dependencies:

  • NodeJS 16+ (or 14 using node --experimental-wasm-threads)
  • NPM 6+
  • Git 2+
tip

If you have an older version installed, we suggest installing a newer version using the package manager for your system: Homebrew (Mac), Chocolatey (Windows), or apt/yum/etc (Linux). On Linux, you may need to install a recent NodeJS version via NodeSource (deb or rpm), as recommended by the NodeJS Project.

Start a project

Now that you have Mina zkApp CLI installed, you can start with an example or start your own project.

Examples are based on the standard project structure, but with additional files in the /src directory as the only difference.

  1. Install: Run zk example sudoku. This creates a new project and includes the example files (i.e. the smart contract) inside the project’s src/ directory. Type ls & hit enter to see the files that were created or open the directory in a code editor such as VS Code.
  2. Run tests: Run npm run test. Tests are written using Jest. After running this command, you should see all tests pass. You can also run npm run testw to run tests in watch mode, so it will automatically re-run tests when saving changes to your code.
  3. Build the example: Run npm run build. This will compile your TypeScript into JavaScript inside the project’s /build directory.
  4. Deploy to Testnet: Run zk config, which will walk you through adding a network alias to your project’s config.json. For Berkeley Testnet, we recommend using berkeley as the name, 0.1 for the fee, and https://proxy.berkeley.minaexplorer.com/graphql for the url. Then run zk deploy and follow the prompts. See the how to deploy a zkApp page for further details.

You can view a list of all available examples here.

Option B: Start your own project

  1. Install: Run zk project <myproj>. Type ls and hit enter to see the newly created project structure.
  2. Run tests: Run npm run test. Tests are written using Jest. After running this command, you should see all tests pass. You can also run npm run testw to run tests in watch mode, so it will automatically re-run tests when saving changes to your code.
  3. Build: Run npm run build. This will compile your TypeScript code into JavaScript inside the project’s /build.
  4. Deploy to Testnet: Run zk config, which will walk you through adding a network alias to your project’s config.json. For Berkeley Testnet, we recommend using berkeley as the name, 0.1 for the fee, and https://proxy.berkeley.minaexplorer.com/graphql for the url. Then run zk deploy and follow the prompts. See the how to deploy a zkApp page for further details.
  5. Deploy to Mainnet: (Coming soon.)

Writing your smart contract

The goal of this section is to explain the concepts that you will need to understand to write a zero-knowledge-based smart contract.

If you haven’t yet read the how zkApps work pages, please read it first so that this section makes sense.

SnarkyJS

zkApps are written in TypeScript using SnarkyJS. SnarkyJS is a TypeScript library for writing smart contracts based on zero-knowledge proofs for the Mina Protocol. It is included automatically when creating a new project using the Mina zkApp CLI.

To view the full SnarkyJS reference, please see the snarkyJS reference.

Concepts

Field elements are the basic unit of data in zero-knowledge proof programming. Each field element can store a number up to almost 256 bits in size. You can think of it as a uint256 in Solidity.

note

For the cryptography inclined, the exact max value that a field can store is: 28,948,022,309,329,048,855,892,746,252,171,976,963,363,056,481,941,560,715,954,676,764,349,967,630,336

For example, in typical programming, you might use:

const sum = 1 + 3.

In SnarkyJS, you would write this as:

const sum = new Field(1).add(new Field(3))

This can be simplified as:

const sum = new Field(1).add(3)

Note that the 3 is auto-promoted to a field type to make this cleaner.

Primitive data types

A couple common data types you may use are:

new Bool(x);   // accepts true or false
new Field(x); // accepts an integer, or a numeric string if you want to represent a number greater than JavaScript can represent but within the max value that a field can store.
new UInt64(x); // accepts a Field - useful for constraining numbers to 64 bits
new UInt32(x); // accepts a Field - useful for constraining numbers to 32 bits

PrivateKey, PublicKey, Signature; // useful for accounts and signing
new Group(x, y); // a point on our elliptic curve, accepts two Fields/numbers/strings
Scalar; // the corresponding scalar field (different than Field)

CircuitString.from('some string'); // string of max length 128

In the case of Field and Bool, you can also call the constructor without new:

let x = Field(10);
let b = Bool(true);
Conditionals

Traditional conditional statements are not yet supported by SnarkyJS:

// this will NOT work
if (foo) {
x.assertEquals(y);
}

Instead, use SnarkyJS’ built-in Circuit.if() method, which is a ternary operator:

const x = Circuit.if(new Bool(foo), a, b); // behaves like `foo ? a : b`
Functions

Functions work as you would expect in TypeScript. For example:

function addOneAndDouble(x: Field): Field {
return x.add(1).mul(2);
}
Common methods

Some common methods you will use often are:

let x = new Field(4); // x = 4
x = x.add(3); // x = 7
x = x.sub(1); // x = 6
x = x.mul(3); // x = 18
x = x.div(2); // x = 9
x = x.square(); // x = 81
x = x.sqrt(); // x = 9

let b = x.equals(8); // b = Bool(false)
b = x.greaterThan(8); // b = Bool(true)
b = b.not().or(b).and(b); // b = Bool(true)
b.toBoolean(); // true

let hash = Poseidon.hash([x]); // takes array of Fields, returns Field

let privKey = PrivateKey.random(); // create a private key
let pubKey = PublicKey.fromPrivateKey(privKey); // derive public key
let msg = [hash];
let sig = Signature.create(privKey, msg); // sign a message
sig.verify(pubKey, msg); // Bool(true)

For a full list, see the SnarkyJS reference.

Smart Contract

Now that we have covered the basics of writing SnarkyJS programs, let's see how to create a smart contract.

Smart contracts are written by extending the base class SmartContract:

class HelloWorld extends SmartContract {}

The constructor of a SmartContract is inherited from the base class and should not be overriden. It takes the zkApp account address (a public key) as its only argument:

let zkAppKey = PrivateKey.random();
let zkAppAddress = PublicKey.fromPrivateKey(zkAppKey);

let zkApp = new HelloWorld(zkAppAddress);

Later, we show you how you can deploy a smart contract to an on-chain account.

note

On Mina, there is no strong distinction between normal "user accounts" and "zkApp accounts". A zkApp account is just a normal account that has a smart contract deployed to it -- which essentially just means there's a verification key stored on the account, which can verify zero-knowledge proofs generated with the smart contract.

Methods

Interaction with a smart contract happens by calling one or more of its methods. You declare methods using the @method decorator:

class HelloWorld extends SmartContract {
@method myMethod(x: Field) {
x.mul(2).assertEquals(5);
}
}

Within a method, you can use SnarkyJS' data types and methods to define your custom logic.

Later, we'll show how you can...

  • run a method (off-chain)
  • create a proof that it executed successfully
  • send that proof to the Mina network, to trigger actions like a state change or payment

To get an idea what "successful execution" means, look at this line in our example above:

x.mul(2).assertEquals(5);

Creating a proof for this method will only be possible if the input x satisfies the equation x * 2 === 5. This is what we call a "constraint". Magically, the proof can be checked without seeing x -- it's a private input.

The method above is not very meaningful yet. To make it more interesting, we need a way to interact with accounts, and record state on-chain. Check out the next section for more on that!

One more note about private inputs: The method above has one input parameter, x of type Field. In general, arguments can be any of the built-in SnarkyJS type that you saw: Bool, UInt64, PrivateKey, etc. From now on, we will refer to those types as structs`.

info

Under the hood, every @method defines a zk-SNARK circuit. From the cryptography standpoint, a smart contract is a collection of circuits, all of which are compiled into a single prover & verification key. The proof says something to the effect of "I ran one of these methods, with some private input, and it produced this particular set of account updates". In ZKP terms, the account updates are the public input. The proof will only be accepted on the network if it verifies against the verification key stored in the account. This ensures that indeed, the same code that the zkApp developer wrote also ran on the user's device -- thus, the account updates conform to the smart contract's rules.

tip

You will find that inside a @method, things sometimes behave a little differently. For example, the following code can't be used in a method where x: Field is an input parameter:

console.log(x.toString()); // don't do this inside a `@method`! 😬

This doesn't work because, when we compile the SmartContract into prover and verification keys, we will run your method in an environment where the method inputs don't have any concrete values attached to them. They are like mathematical variables x, y, z which are used to build up abstract computations like x^2 + y^2, just by running your method code.

Therefore, when executing your code and trying to read the value of x to turn it into a string via x.toString(), it will blow up because such a value can't be found. On the other hand, during proof generation all the variables have actual values attached to them (cryptographers call them "witnesses"); and it makes perfect sense to want to log these values for debugging. This is why we have a special function for logging stuff from inside your method:

Circuit.log(x);

The API is like that of console.log, but it will automatically handle printing SnarkyJS data types in a nice format. During SmartContract compilation, it will simply do nothing.

On-chain state

A smart contract can contain on-chain state, which is declared as a property on the class with the @state decorator:

class HelloWorld extends SmartContract {
@state(Field) x = State<Field>();

// ...
}

Here, x is of type Field. Like with method inputs, only SnarkyJS structs can be used for state variables. In the current design, the state can consist of at most 8 Fields of 32 bytes each. These states are stored on the zkApp account. Some structs take up more than one Field: for example, a PublicKey needs 2 of the 8 Fields. States are initialized with the State() function.

A method can modify on-chain state by using this.<state>.set():

class HelloWorld extends SmartContract {
@state(Field) x = State<Field>();

@method setX(x: Field) {
this.x.set(x);
}
}

As a zkApp developer, if you add this method to your smart contract, you are saying: "Anyone can call this method, to set x on the account to any value they want."

Reading state

Often, we also want to read state -- check out this example:

class HelloWorld extends SmartContract {
@state(Field) x = State<Field>();

@method increment() {
// read state
const x = this.x.get();
this.x.assertEquals(x);

// write state
this.x.set(x.add(1));
}
}

The increment() method fetches the current on-chain state x with this.x.get(). Later, it sets the new state to x + 1 using this.x.set(). Simple!

There's another line though, which looks weird at first:

this.x.assertEquals(x);

To understand it, we have to take a step back, and understand what it means to "use an on-chain value" during off-chain execution.

For sure, when we use an on-chain value, we have to prove that this is the on-chain value. Verification has to fail if it's a different value! Otherwise, a malicious user could modify SnarkyJS and make it just use any other value than the current on-chain state -- breaking our zkApp.

To prevent that, we link "x at proving time" to be the same as "x at verification time". We call this a precondition -- a condition that is checked by the verifier (a Mina node) when it receives the proof in a transaction. This is what this.x.assertEquals(x) does: it adds the precondition that this.x -- the on-chain state at verification time -- has to equal x -- the value we fetched from the chain on the client-side. In zkSNARK language, x becomes part of the public input.

Side note: this.<state>.assertEquals is more flexible than equating with the current value. For example, this.x.assertEquals(10) fixes the on-chain x to the number 10.

note

Why didn't we just make this.x.get() add the precondition, automatically, so that you didn't have to write this.x.assertEquals(x)? Well, we like to keep things explicit. The assertion reminds us that we add logic which can make the proof fail: If x isn't the same at verification time, the transaction will be rejected. So, reading on-chain values has to be done with care if many users are supposed to read and update state concurrently. It is applicable in some situations, but might cause races, and call for workarounds, in other situations. One such workaround is the use of actions -- see Actions and Reducer.

Assertions

Let's modify the increment() method to accept a parameter:

class HelloWorld extends SmartContract {
@state(Field) x = State<Field>();

@method increment(xPlus1: Field) {
const x = this.x.get();
this.x.assertEquals(x);

x.add(1).assertEquals(xPlus1);

this.x.set(xPlus1);
}
}

Here, after obtaining the current state x and asserting that it equals the on-chain value, we make another assertion:

x.add(1).assertEquals(xPlus1);

If the assertion fails, SnarkyJS will throw an error and not submit the transaction. On the other hand, if it succeeds, it becomes part of the proof that is verified on-chain.

Because of this, our new version of increment() is guaranteed to behave like the previous version: It can only ever update the state x to x + 1.

tip

You can add optional failure messages to assertions, to make debugging easier. For example, the above example could be written as:

x.add(1).assertEquals(xPlus1, 'x + 1 should equal xPlus1');

Assertions can be incredibly useful to constrain state updates. Common assertions you may use are:

x.assertEquals(y); // x = y
x.assertBoolean(); // x = 0 or x = 1
x.assertLt(y); // x < y
x.assertLte(y); // x <= y
x.assertGt(y); // x > y
x.assertGte(y); // x >= y

For a full list, see the SnarkyJS reference.

Public and private inputs

We touched on this already, but it's good to cover it in more depth:

While the state of a zkApp is public, method parameters are private.

When a smart contract method is called, the proof it produces uses zero-knowledge to hide inputs and details of the computation.

The only way method parameters can be exposed is when the computation explicitly exposes them, as in our last example where the input was directly stored in the public state: this.x.set(xPlus1);

Let's show an example where this is not the case, by defining a new method called incrementSecret():

class HelloWorld extends SmartContract {
@state(Field) x = State<Field>();

// ...

@method incrementSecret(secret: Field) {
const x = this.x.get();
this.x.assertEquals(x);

Poseidon.hash(secret).assertEquals(x);
this.x.set(Poseidon.hash(secret.add(1)));
}
}

This time, our input is called secret. We check that the hash of our secret is equal to the current state x. If this is the case, we add 1 to the secret and set x to the hash of that.

When running this successfully, it just proves that the code was run with some input secret whose hash is x, and that the new x will be set to hash(secret + 1). However, the secret itself remains private, because it can't be deduced from its hash.

Initializing state

We didn't cover how to initialize on-chain state yet. This can be done in the init() method.

Like the constructor, init() is predefined on the base SmartContract class. It will be called when you deploy your zkApp with the zkApp CLI, for the first time. It won't be called if you upgrade your contract and deploy a second time. You can override this method to add initialization of your on-chain state:

class HelloWorld extends SmartContract {
@state(Field) x = State<Field>();

init() {
super.init();
this.x.set(Field(10)); // initial state
}
}

Note that we have to call super.init(), which sets your entire state to 0.

If you don't have any state to initialize to values other than 0, then there's no need to override init(), you can just leave it out. In the example above however, we are setting our state x to Field(10).

Custom data types

As mentioned previously, smart contract method arguments can be any of the built-in SnarkyJS types.

However, what if you want to define your own data type?

You can create a custom data type for your smart contract using the Struct function that SnarkyJS exposes. To do this, create a class that extends Struct({ }). Then, inside the object { }, define the fields that you want to use in your custom data type.

For example, let us say you want to create a custom data type called Point representing a 2D point on a grid. The Point struct has no instance methods and is only used to hold information about the x and y points. You can create a such a Point class by creating a new class that extends the Struct class:

class Point extends Struct({
x: Field,
y: Field,
}) {}

Now that you have defined your Struct, you can use it in your smart contract for any SnarkyJS built-in types.

For example, the following smart contract uses the Point Struct defined above as state and as a method argument:

export class Grid extends SmartContract {
@state(Point) p = State<Point>();

@method init() {
this.p.set(new Point({ x: Field(1), y: Field(2) }));
}

@method move(newPoint: Point) {
const point = this.p.get();
this.p.assertEquals(point);

const newX = point.x.add(newPoint.x);
const newY = point.y.add(newPoint.y);

this.p.set(new Point({ x: newX, y: newY }));
}
}

Note that your Structs can contain SnarkyJS built-in types like Field, Bool, UInt64, etc or even other custom types that you've defined which are based on the Struct class. This allows for great composability and reusability of structs.

Next Steps

Now that you've learned how to write a basic smart contract, you can learn how to test your zkApp.