Actions and Blinks

Solana Actions are specification-compliant APIs that return transactions on the Solana blockchain to be previewed, signed, and sent across a number of various contexts, including QR codes, buttons + widgets, and websites across the internet. Actions make it simple for developers to integrate the things you can do throughout the Solana ecosystem right into your environment, allowing you to perform blockchain transactions without needing to navigate away to a different app or webpage.

Blockchain links – or blinks – turn any Solana Action into a shareable, metadata-rich link. Blinks allow Action-aware clients (browser extension wallets, bots) to display additional capabilities for the user. On a website, a blink might immediately trigger a transaction preview in a wallet without going to a decentralized app; in Discord, a bot might expand the blink into an interactive set of buttons. This pushes the ability to interact on-chain to any web surface capable of displaying a URL.

Get Started #

To quickly get started with creating custom Solana Actions:

npm install @solana/actions
  • install the Solana Actions SDK in your application
  • build an API endpoint for the GET request that returns the metadata about your Action
  • create an API endpoint that accepts the POST request and returns the signable transaction for the user
Info

Checkout this video tutorial on how to build a Solana Action using the @solana/actions SDK.

You can also find the source code for an Action that performs a native SOL transfer here and several other example Actions in this repo.

When deploying your custom Solana Actions to production:

If you are looking for inspiration around building Actions and blinks, checkout the Awesome Blinks repository for some community creations and even ideas for new ones.

Actions #

The Solana Actions specification uses a set of standard APIs to deliver signable transactions (and eventually signable messages) from an application directly to a user. They are hosted at publicly accessible URLs and are therefore accessible by their URL for any client to interact with.

Info

You can think of Actions as a API endpoint that will return metadata and something for a user to sign (either a transaction or a authentication message) with their blockchain wallet.

The Actions API consists of making simple GET and POST requests to an Action's URL endpoint and handling the responses that conform to the Actions interface.

  1. the GET request returns metadata that provides human-readable information to the client about what actions are available at this URL, and an optional list of related actions.
  2. the POST request returns a signable transaction or message that the client then prompts the user's wallet to sign and execute on the blockchain or in another offchain service.

Action Execution and Lifecycle #

In practice, interacting with Actions closely resembles interacting with a typical REST API:

  • a client makes the initial GET request to an Action URL in order to fetch metadata about the Actions available
  • the endpoint returns a response that include metadata about the endpoint (like the application's title and icon) and a listing of the available actions for this endpoint
  • the client application (like a mobile wallet, chat bot, or website) displays a UI for the user to perform one of the actions
  • after the user selects an action (by clicking a button), the client makes a POST request to the endpoint in order to get the transaction for the user to sign
  • the wallet facilitates the user signing the transaction and ultimately sends the transaction to the blockchain for confirmation

Solana Actions Execution and LifecycleSolana Actions Execution and Lifecycle

When receiving transactions from an Actions URL, clients should handle submission of these transactions to the blockchain and manage their state lifecycle.

Actions also support some level of invalidation before execution. The GET and POST request may return some metadata that states whether the action is capable of be taken (like with the disabled field).

For example, if there was an Action endpoint that facilitates voting on a DAO governance proposal whose voting window has closed, the initial GET request may return the error message "This proposal is no longer up for a vote" and the "Vote Yes" and "Vote No" buttons as "disabled".

Blinks (blockchain links) are client applications that introspect Action APIs and construct user interfaces around interacting with and executing Actions.

Client applications that support blinks simply detect Action-compatible URLs, parse them, and allow users to interact with them in standardized user interfaces.

Info

Any client application that fully introspects an Actions API to build a complete interface for it is a blink. Therefore, not all clients that consume Actions APIs are blinks.

A blink URL describes a client application that enables a user to complete the full lifecycle of executing an Action, including signing with their wallet.

https://example.domain/?action=<action_url>

For any client application to become a blink:

  • The blink URL must contain a query parameter of action whose value is a URL-encoded Action URL. This value must be URL-encoded to not conflict with any other protocol parameters.

  • The client application must URL-decode the action query parameter and introspect the Action API link provided (see Action URL scheme).

  • The client must render a rich user interface that enables a user to complete the full lifecycle of executing an Action, including signing with their wallet.

Info

Not all blink client applications (e.g. websites or dApps) will support all Actions. Application developers may chose which Actions they want to support within their blink interfaces.

The following example demonstrates a valid blink URL with an action value of solana-action:https://actions.alice.com/donate that is URL encoded:

https://example.domain/?action=solana-action%3Ahttps%3A%2F%2Factions.alice.com%2Fdonate

Blinks may be linked to Actions in at least 3 ways:

  1. Sharing an explicit Action URL: solana-action:https://actions.alice.com/donate

    In this case, only supported clients may render the blink. There will be no fallback link preview, or site that may be visited outside of the non-supporting client.

  2. Sharing a link to a website that is linked to an Actions API via an actions.json file on the website's domain root.

    For example, https://alice.com/actions.json maps https://alice.com/donate, a website URL at which users can donate to Alice, to API URL https://actions.alice.com/donate, at which Actions for donating to Alice are hosted.

  3. Embedding an Action URL in an “interstitial” site URL that understands how to parse Actions.

    https://example.domain/?action=<action_url>

Clients that support blinks should be able to take any of the above formats and correctly render an interface to facilitate executing the action directly in the client.

For clients that do not support blinks, there should be an underlying website (making the browser become the universal fallback).

If a user taps anywhere on a client that is not an action button or text input field, they should be taken to the underlying site.

While Solana Actions and blinks are a permissionless protocol/specification, client applications and wallets are still required to ultimately facilitate users to sign the transaction.

Info

Use the Blinks Inspector tool to inspect, debug, and test your blinks and actions directly in your browser. You can view the GET and POST response payloads, response headers, and test all inputs to each of your linked Actions.

Each client application or wallets may have different requirements on which Action endpoints their clients will automatically unfurl and immediately display to their users on social media platforms.

For example, some clients may operate on an "allow list" approach that may require verification prior to their client unfurling an Action for users such as Dialect's Actions Registry (detailed below).

All blinks will still render and allow for signing on Dialect's dial.to blinks Interstitial site, with their registry status displayed in the blink.

Dialect's Actions Registry #

As a public good for the Solana ecosystem, Dialect maintains a public registry — together with the help of Solana Foundation and other community members — of blockchain links that have are from pre-verified from known sources. As of launch, only Actions that have been registered in the Dialect registry will unfurl in the Twitter feed when posted.

Client applications and wallets can freely choose to use this public registry or another solution to help ensure user security and safety. If not verified through the Dialect registry, the blockchain link will not be touched by the blink client, and will be rendered as a typical URL.

Developers can apply to be verified by Dialect here: dial.to/register

Specification #

The Solana Actions specification consists of key sections that are part of a request/response interaction flow:

Each of these requests are made by the Action client (e.g. wallet app, browser extension, dApp, website, etc) to gather specific metadata for rich user interfaces and to facilitate user input to the Actions API.

Each of the responses are crafted by an application (e.g. website, server backend, etc) and returned to the Action client. Ultimately, providing a signable transaction or message for a wallet to prompt the user to approve, sign, and send to the blockchain.

Info

The types and interfaces declared within this readme files are often the simplified version of the types to aid in readability.

For better type safety and improved developer experience, the @solana/actions-spec package contains more complex type definitions. You can find the source code for them here.

URL Scheme #

A Solana Action URL describes an interactive request for a signable Solana transaction or message using the solana-action protocol.

The request is interactive because the parameters in the URL are used by a client to make a series of standardized HTTP requests to compose a signable transaction or message for the user to sign with their wallet.

solana-action:<link>
  • A single link field is required as the pathname. The value must be a conditionally URL-encoded absolute HTTPS URL.

  • If the URL contains query parameters, it must be URL-encoded. URL-encoding the value prevents conflicting with any Actions protocol parameters, which may be added via the protocol specification.

  • If the URL does not contain query parameters, it should not be URL-encoded. This produces a shorter URL and a less dense QR code.

In either case, clients must URL-decode the value. This has no effect if the value isn't URL-encoded. If the decoded value is not an absolute HTTPS URL, the wallet must reject it as malformed.

OPTIONS response #

In order to allow Cross-Origin Resource Sharing (CORS) within Actions clients (including blinks), all Action endpoints should respond to HTTP requests for the OPTIONS method with valid headers that will allow clients to pass CORS checks for all subsequent requests from their same origin domain.

An Actions client may perform "preflight" requests to the Action URL endpoint in order check if the subsequent GET request to the Action URL will pass all CORS checks. These CORS preflight checks are made using the OPTIONS HTTP method and should respond with all required HTTP headers that will allow Action clients (like blinks) to properly make all subsequent requests from their origin domain.

At a minimum, the required HTTP headers include:

  • Access-Control-Allow-Origin with a value of *
    • this ensures all Action clients can safely pass CORS checks in order to make all required requests
  • Access-Control-Allow-Methods with a value of GET,POST,PUT,OPTIONS
    • ensures all required HTTP request methods are supported for Actions
  • Access-Control-Allow-Headers with a minimum value of Content-Type, Authorization, Content-Encoding, Accept-Encoding

For simplicity, developers should consider returning the same response and headers to OPTIONS requests as their GET response.

Cross-Origin headers for actions.json

The actions.json file response must also return valid Cross-Origin headers for GET and OPTIONS requests, specifically the Access-Control-Allow-Origin header value of *.

See actions.json below for more details.

GET Request #

The Action client (e.g. wallet, browser extension, etc) should make an HTTP GET JSON request to the Action's URL endpoint.

  • The request should not identify the wallet or the user.
  • The client should make the request with an Accept-Encoding header.
  • The client should display the domain of the URL as the request is being made.

GET Response #

The Action's URL endpoint (e.g. application or server backend) should respond with an HTTP OK JSON response (with a valid payload in the body) or an appropriate HTTP error.

Error responses (i.e. HTTP 4xx and 5xx status codes) should return a JSON response body following ActionError to present a helpful error message to users. See Action Errors.

GET Response Body #

A GET response with an HTTP OK JSON response should include a body payload that follows the interface specification:

ActionGetResponse
export type ActionType = "action" | "completed";
 
export type ActionGetResponse = Action<"action">;
 
export interface Action<T extends ActionType> {
  /** type of Action to present to the user */
  type: T;
  /** image url that represents the source of the action request */
  icon: string;
  /** describes the source of the action request */
  title: string;
  /** brief summary of the action to be performed */
  description: string;
  /** button text rendered to the user */
  label: string;
  /** UI state for the button being rendered to the user */
  disabled?: boolean;
  links?: {
    /** list of related Actions a user could perform */
    actions: LinkedAction[];
  };
  /** non-fatal error message to be displayed to the user */
  error?: ActionError;
}
  • type - The type of action being given to the user. Defaults to action. The initial ActionGetResponse is required to have a type of action.

    • action - Standard action that will allow the user to interact with any of the LinkedActions
    • completed - Used to declare the "completed" state within action chaining.
  • icon - The value must be an absolute HTTP or HTTPS URL of an icon image. The file must be an SVG, PNG, or WebP image, or the client/wallet must reject it as malformed.

  • title - The value must be a UTF-8 string that represents the source of the action request. For example, this might be the name of a brand, store, application, or person making the request.

  • description - The value must be a UTF-8 string that provides information on the action. The description should be displayed to the user.

  • label - The value must be a UTF-8 string that will be rendered on a button for the user to click. All labels should not exceed 5 word phrases and should start with a verb to solidify the action you want the user to take. For example, "Mint NFT", "Vote Yes", or "Stake 1 SOL".

  • disabled - The value must be boolean to represent the disabled state of the rendered button (which displays the label string). If no value is provided, disabled should default to false (i.e. enabled by default). For example, if the action endpoint is for a governance vote that has closed, set disabled=true and the label could be "Vote Closed".

  • error - An optional error indication for non-fatal errors. If present, the client should display it to the user. If set, it should not prevent the client from interpreting the action or displaying it to the user (see Action Errors). For example, the error can be used together with disabled to display a reason like business constraints, authorization, the state, or an error of external resource.

  • links.actions - An optional array of related actions for the endpoint. Users should be displayed UI for each of the listed actions and expected to only perform one. For example, a governance vote action endpoint may return three options for the user: "Vote Yes", "Vote No", and "Abstain from Vote".

    • If no links.actions is provided, the client should render a single button using the root label string and make the POST request to the same action URL endpoint as the initial GET request.

    • If any links.actions are provided, the client should only render buttons and input fields based on the items listed in the links.actions field. The client should not render a button for the contents of the root label.

LinkedAction
export interface LinkedAction {
  /** URL endpoint for an action */
  href: string;
  /** button text rendered to the user */
  label: string;
  /**
   * Parameters to accept user input within an action
   * @see {ActionParameter}
   * @see {ActionParameterSelectable}
   */
  parameters?: Array<TypedActionParameter>;
}

The ActionParameter allows declaring what input the Action API is requesting from the user:

ActionParameter
/**
 * Parameter to accept user input within an action
 * note: for ease of reading, this is a simplified type of the actual
 */
export interface ActionParameter {
  /** input field type */
  type?: ActionParameterType;
  /** parameter name in url */
  name: string;
  /** placeholder text for the user input field */
  label?: string;
  /** declare if this field is required (defaults to `false`) */
  required?: boolean;
  /** regular expression pattern to validate user input client side */
  pattern?: string;
  /** human-readable description of the `type` and/or `pattern`, represents a caption and error, if value doesn't match */
  patternDescription?: string;
  /** the minimum value allowed based on the `type` */
  min?: string | number;
  /** the maximum value allowed based on the `type` */
  max?: string | number;
}

The pattern should be a string equivalent of a valid regular expression. This regular expression pattern should by used by blink-clients to validate user input before before making the POST request. If the pattern is not a valid regular expression, it should be ignored by clients.

The patternDescription is a human readable description of the expected input requests from the user. If pattern is provided, the patternDescription is required to be provided.

The min and max values allows the input to set a lower and/or upper bounds of the input requested from the user (i.e. min/max number and or min/max character length), and should be used for client side validation. For input types of date or datetime-local, these values should be a string dates. For other string based input types, the values should be numbers representing their min/max character length.

If the user input value is not considered valid per the pattern, the user should receive a client side error message indicating the input field is not valid and displayed the patternDescription string.

The type field allows the Action API to declare more specific user input fields, providing better client side validation and improving the user experience. In many cases, this type will resemble the standard HTML input element.

The ActionParameterType can be simplified to the following type:

ActionParameterType
/**
 * Input field type to present to the user
 * @default `text`
 */
export type ActionParameterType =
  | "text"
  | "email"
  | "url"
  | "number"
  | "date"
  | "datetime-local"
  | "checkbox"
  | "radio"
  | "textarea"
  | "select";

Each of the type values should normally result in a user input field that resembles a standard HTML input element of the corresponding type (i.e. <input type="email" />) to provide better client side validation and user experience:

  • text - equivalent of HTML “text” input element
  • email - equivalent of HTML “email” input element
  • url - equivalent of HTML “url” input element
  • number - equivalent of HTML “number” input element
  • date - equivalent of HTML “date” input element
  • datetime-local - equivalent of HTML “datetime-local” input element
  • checkbox - equivalent to a grouping of standard HTML “checkbox” input elements. The Action API should return options as detailed below. The user should be able to select multiple of the provided checkbox options.
  • radio - equivalent to a grouping of standard HTML “radio” input elements. The Action API should return options as detailed below. The user should be able to select only one of the provided radio options.
  • Other HTML input type equivalents not specified above (hidden, button, submit, file, etc) are not supported at this time.

In addition to the elements resembling HTML input types above, the following user input elements are also supported:

  • textarea - equivalent of HTML textarea element. Allowing the user provide multi-line input.
  • select - equivalent of HTML select element, allowing the user to experience a “dropdown” style field. The Action API should return options as detailed below.

When type is set as select, checkbox, or radio then the Action API should include an array of options that each provide a label and value at a minimum. Each option may also have a selected value to inform the blink-client which of the options should be selected by default for the user (see checkbox and radio for differences).

This ActionParameterSelectable can be simplified to the following type definition:

ActionParameterSelectable
/**
 * note: for ease of reading, this is a simplified type of the actual
 */
interface ActionParameterSelectable extends ActionParameter {
  options: Array<{
    /** displayed UI label of this selectable option */
    label: string;
    /** value of this selectable option */
    value: string;
    /** whether or not this option should be selected by default */
    selected?: boolean;
  }>;
}

If no type is set or an unknown/unsupported value is set, blink-clients should default to text and render a simple text input.

The Action API is still responsible to validate and sanitize all data from the user input parameters, enforcing any “required” user input as necessary.

For platforms other that HTML/web based ones (like native mobile), the equivalent native user input component should be used to achieve the equivalent experience and client side validation as the HTML/web input types described above.

Example GET Response #

The following example response provides a single "root" action that is expected to be presented to the user a single button with a label of "Claim Access Token":

{
  "title": "HackerHouse Events",
  "icon": "<url-to-image>",
  "description": "Claim your Hackerhouse access token.",
  "label": "Claim Access Token" // button text
}

The following example response provides 3 related action links that allow the user to click one of 3 buttons to cast their vote for a DAO proposal:

{
  "title": "Realms DAO Platform",
  "icon": "<url-to-image>",
  "description": "Vote on DAO governance proposals #1234.",
  "label": "Vote",
  "links": {
    "actions": [
      {
        "label": "Vote Yes", // button text
        "href": "/api/proposal/1234/vote?choice=yes"
      },
      {
        "label": "Vote No", // button text
        "href": "/api/proposal/1234/vote?choice=no"
      },
      {
        "label": "Abstain from Vote", // button text
        "href": "/api/proposal/1234/vote?choice=abstain"
      }
    ]
  }
}

Example GET Response with Parameters #

The following examples response demonstrate how to accept text input from the user (via parameters) and include that input in the final POST request endpoint (via the href field within a LinkedAction):

The following example response provides the user with 3 linked actions to stake SOL: a button labeled "Stake 1 SOL", another button labeled "Stake 5 SOL", and a text input field that allows the user to enter a specific "amount" value that will be sent to the Action API:

{
  "title": "Stake-o-matic",
  "icon": "<url-to-image>",
  "description": "Stake SOL to help secure the Solana network.",
  "label": "Stake SOL", // not displayed since `links.actions` are provided
  "links": {
    "actions": [
      {
        "label": "Stake 1 SOL", // button text
        "href": "/api/stake?amount=1"
        // no `parameters` therefore not a text input field
      },
      {
        "label": "Stake 5 SOL", // button text
        "href": "/api/stake?amount=5"
        // no `parameters` therefore not a text input field
      },
      {
        "label": "Stake", // button text
        "href": "/api/stake?amount={amount}",
        "parameters": [
          {
            "name": "amount", // field name
            "label": "SOL amount" // text input placeholder
          }
        ]
      }
    ]
  }
}

The following example response provides a single input field for the user to enter an amount which is sent with the POST request (either as a query parameter or a subpath can be used):

{
  "icon": "<url-to-image>",
  "label": "Donate SOL",
  "title": "Donate to GoodCause Charity",
  "description": "Help support this charity by donating SOL.",
  "links": {
    "actions": [
      {
        "label": "Donate", // button text
        "href": "/api/donate/{amount}", // or /api/donate?amount={amount}
        "parameters": [
          // {amount} input field
          {
            "name": "amount", // input field name
            "label": "SOL amount" // text input placeholder
          }
        ]
      }
    ]
  }
}

POST Request #

The client must make an HTTP POST JSON request to the action URL with a body payload of:

{
  "account": "<account>"
}
  • account - The value must be the base58-encoded public key of an account that may sign the transaction.

The client should make the request with an Accept-Encoding header and the application may respond with a Content-Encoding header for HTTP compression.

The client should display the domain of the action URL as the request is being made. If a GET request was made, the client should also display the title and render the icon image from that GET response.

POST Response #

The Action's POST endpoint should respond with an HTTP OK JSON response (with a valid payload in the body) or an appropriate HTTP error.

Error responses (i.e. HTTP 4xx and 5xx status codes) should return a JSON response body following ActionError to present a helpful error message to users. See Action Errors.

POST Response Body #

A POST response with an HTTP OK JSON response should include a body payload of:

ActionPostResponse
/**
 * Response body payload returned from the Action POST Request
 */
export interface ActionPostResponse<T extends ActionType = ActionType> {
  /** base64 encoded serialized transaction */
  transaction: string;
  /** describes the nature of the transaction */
  message?: string;
  links?: {
    /**
     * The next action in a successive chain of actions to be obtained after
     * the previous was successful.
     */
    next: NextActionLink;
  };
}
  • transaction - The value must be a base64-encoded serialized transaction. The client must base64-decode the transaction and deserialize it.

  • message - The value must be a UTF-8 string that describes the nature of the transaction included in the response. The client should display this value to the user. For example, this might be the name of an item being purchased, a discount applied to a purchase, or a thank you note.

  • links.next - An optional value use to "chain" multiple Actions together in series. After the included transaction has been confirmed on-chain, the client can fetch and render the next action. See Action Chaining for more details.

  • The client and application should allow additional fields in the request body and response body, which may be added by future specification updates.

Info

The application may respond with a partially or fully signed transaction. The client and wallet must validate the transaction as untrusted.

POST Response - Transaction #

If the transaction signatures are empty or the transaction has NOT been partially signed:

  • The client must ignore the feePayer in the transaction and set the feePayer to the account in the request.
  • The client must ignore the recentBlockhash in the transaction and set the recentBlockhash to the latest blockhash.
  • The client must serialize and deserialize the transaction before signing it. This ensures consistent ordering of the account keys, as a workaround for this issue.

If the transaction has been partially signed:

  • The client must NOT alter the feePayer or recentBlockhash as this would invalidate any existing signatures.
  • The client must verify existing signatures, and if any are invalid, the client must reject the transaction as malformed.

The client must only sign the transaction with the account in the request, and must do so only if a signature for the account in the request is expected.

If any signature except a signature for the account in the request is expected, the client must reject the transaction as malicious.

Action Errors #

Actions APIs should return errors using ActionError in order to present helpful error messages to the user. Depending on the context, this error could be fatal or non-fatal.

ActionError
export interface ActionError {
  /** simple error message to be displayed to the user */
  message: string;
}

When an Actions API responds with an HTTP error status code (i.e. 4xx and 5xx), the response body should be a JSON payload following ActionError. The error is considered fatal and the included message should be presented to the user.

For API responses that support the optional error attribute (like ActionGetResponse), the error is considered non-fatal and the included message should be presented to the user.

Action Chaining #

Solana Actions can be "chained" together in a successive series. After an Action's transaction is confirmed on-chain, the next action can be obtained and presented to the user.

Action chaining allows developers to build more complex and dynamic experiences within blinks, including:

  • providing multiple transactions (and eventually sign message) to a user
  • customized action metadata based on the user's wallet address
  • refreshing the blink metadata after a successful transaction
  • receive an API callback with the transaction signature for additional validation and logic on the Action API server
  • customized "success" messages by updating the displayed metadata (e.g. a new image and description)

To chain multiple actions together, in any ActionPostResponse include a links.next of either:

  • PostNextActionLink - POST request link with a same origin callback url to receive the signature and user's account in the body. This callback url should respond with a NextAction.
  • InlineNextActionLink - Inline metadata for the next action to be presented to the user immediately after the transaction has confirmed. No callback will be made.
export type NextActionLink = PostNextActionLink | InlineNextActionLink;
 
/** @see {NextActionPostRequest} */
export interface PostNextActionLink {
  /** Indicates the type of the link. */
  type: "post";
  /** Relative or same origin URL to which the POST request should be made. */
  href: string;
}
 
/**
 * Represents an inline next action embedded within the current context.
 */
export interface InlineNextActionLink {
  /** Indicates the type of the link. */
  type: "inline";
  /** The next action to be performed */
  action: NextAction;
}

NextAction #

After the ActionPostResponse included transaction is signed by the user and confirmed on-chain, the blink client should either:

  • execute the callback request to fetch and display the NextAction, or
  • if a NextAction is already provided via links.next, the blink client should update the displayed metadata and make no callback request

If the callback url is not the same origin as the initial POST request, no callback request should be made. Blink clients should display an error notifying the user.

NextAction
/** The next action to be performed */
export type NextAction = Action<"action"> | CompletedAction;
 
/** The completed action, used to declare the "completed" state within action chaining. */
export type CompletedAction = Omit<Action<"completed">, "links">;

Based on the type, the next action should be presented to the user via blink clients in one of the following ways:

  • action - (default) A standard action that will allow the user to see the included Action metadata, interact with the provided LinkedActions, and continue to chain any following actions.

  • completed - The terminal state of an action chain that can update the blink UI with the included Action metadata, but will not allow the user to execute further actions.

If no links.next is not provided, blink clients should assume the current action is final action in the chain, presenting their "completed" UI state after the transaction is confirmed.

actions.json #

The purpose of the actions.json file allows an application to instruct clients on what website URLs support Solana Actions and provide a mapping that can be used to perform GET requests to an Actions API server.

Cross-Origin headers are required

The actions.json file response must also return valid Cross-Origin headers for GET and OPTIONS requests, specifically the Access-Control-Allow-Origin header value of *.

See OPTIONS response above for more details.

The actions.json file should be stored and universally accessible at the root of the domain.

For example, if your web application is deployed to my-site.com then the actions.json file should be accessible at https://my-site.com/actions.json. This file should also be Cross-Origin accessible via any browser by having a Access-Control-Allow-Origin header value of *.

Rules #

The rules field allows the application to map a set of a website's relative route paths to a set of other paths.

Type: Array of ActionRuleObject.

ActionRuleObject
interface ActionRuleObject {
  /** relative (preferred) or absolute path to perform the rule mapping from */
  pathPattern: string;
  /** relative (preferred) or absolute path that supports Action requests */
  apiPath: string;
}
  • pathPattern - A pattern that matches each incoming pathname.

  • apiPath - A location destination defined as an absolute pathname or external URL.

Rules - pathPattern #

A pattern that matches each incoming pathname. It can be an absolute or relative path and supports the following formats:

  • Exact Match: Matches the exact URL path.

    • Example: /exact-path
    • Example: https://website.com/exact-path
  • Wildcard Match: Uses wildcards to match any sequence of characters in the URL path. This can match single (using *) or multiple segments (using **). (see Path Matching below).

    • Example: /trade/* will match /trade/123 and /trade/abc, capturing only the first segment after /trade/.
    • Example: /category/*/item/** will match /category/123/item/456 and /category/abc/item/def.
    • Example: /api/actions/trade/*/confirm will match /api/actions/trade/123/confirm.

Rules - apiPath #

The destination path for the action request. It can be defined as an absolute pathname or an external URL.

  • Example: /api/exact-path
  • Example: https://api.example.com/v1/donate/*
  • Example: /api/category/*/item/*
  • Example: /api/swap/**

Rules - Query Parameters #

Query parameters from the original URL are always preserved and appended to the mapped URL.

Rules - Path Matching #

The following table outlines the syntax for path matching patterns:

OperatorMatches
*A single path segment, not including the surrounding path separator / characters.
**Matches zero or more characters, including any path separator / characters between multiple path segments. If other operators are included, the ** operator must be the last operator.
?Unsupported pattern.

Rules Examples #

The following example demonstrates an exact match rule to map requests requests to /buy from your site's root to the exact path /api/buy relative to your site's root:

actions.json
{
  "rules": [
    {
      "pathPattern": "/buy",
      "apiPath": "/api/buy"
    }
  ]
}

The following example uses wildcard path matching to map requests to any path (excluding subdirectories) under /actions/ from your site's root to a corresponding path under /api/actions/ relative to your site's root:

actions.json
{
  "rules": [
    {
      "pathPattern": "/actions/*",
      "apiPath": "/api/actions/*"
    }
  ]
}

The following example uses wildcard path matching to map requests to any path (excluding subdirectories) under /donate/ from your site's root to a corresponding absolute path https://api.dialect.com/api/v1/donate/ on an external site:

actions.json
{
  "rules": [
    {
      "pathPattern": "/donate/*",
      "apiPath": "https://api.dialect.com/api/v1/donate/*"
    }
  ]
}

The following example uses wildcard path matching for an idempotent rule to map requests to any path (including subdirectories) under /api/actions/ from your site's root to itself:

Info

Idempotent rules allow blink clients to more easily determine if a given path supports Action API requests without having to be prefixed with the solana-action: URI or performing additional response testing.

actions.json
{
  "rules": [
    {
      "pathPattern": "/api/actions/**",
      "apiPath": "/api/actions/**"
    }
  ]
}

Action Identity #

Action endpoints may include an Action Identity in the transactions that are returned in their POST response for the user to sign. This allows indexers and analytics platforms to easily and verifiably attribute on-chain activity to a specific Action Provider (i.e. service) in a verifiable way.

The Action Identity is a keypair used to sign a specially formatted message that is included in transaction using a Memo instruction. This Identifier Message can be verifiably attributed to a specific Action Identity, and therefore attribute transactions to a specific Action Provider.

The keypair is not required to sign the transaction itself. This allows wallets and applications to improve transaction deliverability when no other signatures are on the transaction returned to a user (see POST response transaction).

If an Action Provider's use case requires their backend services to pre-sign the transaction before the user does, they should use this keypair as their Action Identity. This will allow one less account be included in the transaction, lowering the total transactions size by 32-bytes.

Action Identifier Message #

The Action Identifier Message is a colon separate UTF-8 string included in a transaction using a single SPL Memo instruction.

protocol:identity:reference:signature
  • protocol - The value of the protocol being used (set to solana-action per the URL Scheme above)
  • identity - The value must be the base58-encoded public key address of the Action Identity keypair
  • reference - The value must be base58-encoded 32-byte array. This may or may not be public keys, on or off the curve, and may or may not correspond with accounts on Solana.
  • signature - base58-encoded signature created from the Action Identity keypair signing only the reference value.

The reference value must be used only once and in a single transaction. For the purpose of associating transactions with an Action Provider, only the first usage of the reference value is considered valid.

Transactions may have multiple Memo instructions. When performing a getSignaturesForAddress, the results memo field will return each memo instruction's message as a single string with each separated by a semi-colon.

No other data should be included with Identifier Message's Memo instruction.

The identity and the reference should be included as read-only, non-signer keys in the transaction on an instruction that is NOT the Identifier Message Memo instruction.

The Identifier Message Memo instruction must have zero accounts provided. If any accounts are provided, the Memo program requires theses accounts to be valid signers. For the purposes of identifying actions, this restricts flexibility and can degrade the user experience. Therefore it is considered an anti-pattern and must be avoided.

Action Identity Verification #

Any transaction that includes the identity account can be verifiably associated with the Action Provider in a multi-step process:

  1. Get all the transactions for a given identity.
  2. Parse and verify each transaction's memo string, ensuring the signature is valid for the reference stored.
  3. Verify the specific transaction is the first on-chain occurrence of the reference on-chain:
    • If this transaction is the first occurrence, the transaction is considered verified and can a safely attributed to the Action Provider.
    • If this transaction is NOT the first occurrence, it is considered invalid and therefore not attributed to the Action Provider.

Because Solana validators index transactions by the account keys, the getSignaturesForAddress RPC method can be used locate all transactions including the identity account.

This RPC method's response includes all the Memo data in the memo field. If multiple Memo instructions were used in the transaction, each memo message will be included in this memo field and must be parsed accordingly by the verifier to obtain the Identity Verification Message.

These transactions should be initially considered UNVERIFIED. This is due to the identity not being required to sign the transaction which allows any transaction to include this account as a non-signer. Potentially artificially inflating attribution and usage counts.

The Identity Verification Message should be checked to ensure the signature was created by the identity signing the reference. If this signature verification fails, the transaction is invalid and should be attributed to the Action Provider.

If the signature verification is successful, the verifier should ensure this transaction is the first on-chain occurrence of the reference. If it is not, the transaction is considered invalid.