Design Tokens Format Module

Draft Community Group Report

Latest published version:
https://tr.designtokens.org/format/
Editors:
Daniel Banks
Donna Vitan
James Nash
Kevin Powell
Louis Chenais
Feedback:
GitHub design-tokens/community-group (pull requests, new issue, open issues)

Abstract

This document describes the technical specification for a file format to exchange design tokens between different tools.

Status of This Document

This specification was published by the Design Tokens Community Group. It is not a W3C Standard nor is it on the W3C Standards Track. Please note that under the W3C Community Contributor License Agreement (CLA) there is a limited opt-out and other conditions apply. Learn more about W3C Community and Business Groups.

This is a snapshot of the editors' draft. It is provided for discussion only and may change at any moment. Its publication here does not imply endorsement of its contents by W3C or the Design Tokens Community Group Membership. Don't cite this document other than as work in progress.

This document has been published to facilitate Wide Review.

This document was produced by the Design Tokens Community Group, and contributions to this draft are governed by Community Contributor License Agreement (CLA), as specified by the W3C Community Group Process.

GitHub Issues are preferred for discussion of this specification.

1. Conformance

As well as sections marked as non-normative, all authoring guidelines, diagrams, examples, and notes in this specification are non-normative. Everything else in this specification is normative.

The key words MAY, MUST, MUST NOT, SHOULD, and SHOULD NOT in this document are to be interpreted as described in BCP 14 [RFC2119] [RFC8174] when, and only when, they appear in all capitals, as shown here.

2. Introduction

This section is non-normative.

Design tokens are a methodology for expressing design decisions in a platform-agnostic way so that they can be shared across different disciplines, tools, and technologies. They help establish a common vocabulary across organisations.

There is a growing ecosystem of tools for design system maintainers and consumers that incorporate design token functionality, or would benefit from doing so:

It is often desirable for design system teams to integrate such tools together, so that design token data can flow between design and development tools.

For example:

While many tools now offer APIs to access design tokens or the ability to export design tokens as a file, these are all tool-specific. The burden is therefore on design system teams to create and maintain their own, bespoke "glue" code or workflows. Furthermore, if teams want to migrate to different tools, they will need to update those integrations.

This specification aims to facilitate better interoperability between tools and thus lower the work design system teams need to do to integrate them by defining a standard file format for expressing design token data.

3. Terminology

These definitions are focused on the technical aspects of the specification, aimed at implementers such as design tool vendors. Definitions for designers and developers are available at designtokens.org.

3.1 (Design) Token

A (Design) Token is an information associated with a name, at minimum a name/value pair.

For example:

The name may be associated with additional Token Properties.

3.2 (Design) Token Properties

Information associated with a token name.

For example:

Additional metadata may be added by tools and design systems to extend the format as needed.

3.3 Design tool

A design tool is a tool for visual design creation and editing.

For example:

3.4 Translation tool

Design token translation tools translate token data from one format to another.

For example:

3.5 Documentation tool

A documentation tool is a tool for documenting design tokens usage.

For example:

3.6 Type

A token's type is a predefined categorization applied to the token's value.

For example:

Token tools can use Types to infer the purpose of a token.

For example:

3.7 Group

A group is a set of tokens belonging to a specific category.

For example:

Groups are arbitrary and tools SHOULD NOT use them to infer the type or purpose of design tokens.

3.8 Alias (Reference)

A design token's value can be a reference to another token. The same value can have multiple names or aliases.

The following Sass example illustrates this concept:

$color-palette-black: #000000;
$color-text-primary: $color-palette-black;

The value of $color-text-primary is #000000, because $color-text-primary references $color-palette-black. We can also say $color-text-primary is an alias for $color-palette-black.

3.9 Composite (Design) Token

A design token whose value is made up of multiple, named child values. Composite tokens are useful for closely related style properties that are always applied together. For example, a typography style might be made up of a font name, font size, line height, and color.

Here's an example of a composite shadow token:

{
  "shadow-token": {
    "$type": "shadow",
    "$value": {
      "color": "#00000080",
      "offsetX": { "value": 0.5, "unit": "rem" },
      "offsetY": { "value": 0.5, "unit": "rem" },
      "blur": { "value": 1.5, "unit": "rem" },
      "spread": { "value": 0, "unit": "rem" }
    }
  }
}

4. File format

Design token files are JSON (https://www.json.org/) files that adhere to the structure described in this specification.

JSON was chosen as an interchange format on the basis of:

4.1 Media type (MIME type)

When serving design token files via HTTP / HTTPS or in any other scenario where a media type (formerly known as MIME type) needs to be specified, the following MIME type SHOULD be used for design token files:

However, since every design token file is a valid JSON file, they MAY be served using the JSON media type: application/json. The above, more specific media type is preferred and SHOULD be used wherever possible.

Tools that can open design token files MUST support both media types.

4.2 File extensions

When saving design token files on a local file system, it can be useful to have a distinct file extension as this makes them easier to spot in file browsers. It may also help to associate a file icon and a preferred application for opening those files. The following file extensions are recommended by this spec:

The former is more succinct. However, until this format is widely adopted and supported, the latter might be useful to make design token files open in users' preferred JSON editors.

Tools that can open design token files MAY filter available files (e.g. in an open file dialog) to only show ones using those extensions. It is recommended to also provide users with a way of opening files that do not use those extensions (e.g. a "show all files" option or similar).

Tools that can save design token files SHOULD append one of the recommended file extensions to the filename when saving.

Editor's note: JSON schema

The group is currently exploring the addition of a JSON Schema to support the spec.

Editor's note: JSON file size limitations

A concern about file size limitations of JSON files was raised by one of the vendors. The working group continues to gather feedback about any limitations the JSON format imposes.

5. Design token

5.1 Name and value

An object with a $value property is a token. Thus, $value is a reserved word in our spec, meaning you can't have a token whose name is "$value". The parent object's key is the token name.

The example above therefore defines 1 design token with the following properties:

Name and value are both required.

Token names are case-sensitive, so the following example with 2 tokens in the same group whose names only differ in case is valid:

However, some tools MAY need to transform names when exporting to other languages or displaying names to the user, so having token names that differ only in case is likely to cause identical and undesirable duplicates in the output. For example, a translation tool that converts these tokens to Sass code would generate problematic output like this:

Tools MAY display a warning when token names differ only by case.

5.1.1 Character restrictions

All properties defined by this format are prefixed with the dollar sign ($). This convention will also be used for any new properties introduced by future versions of this spec. Therefore, token and group names MUST NOT begin with the $ character.

Furthermore, due to the syntax used for token aliases the following characters MUST NOT be used anywhere in a token or group name:

  • { (left curly bracket)
  • } (right curly bracket)
  • . (period)
Editor's note: '$' Prefix Rationale

Because of the decision to prefix group properties with a dollar sign ($), token properties will also use a dollar sign prefix. This provides a consistent syntax across the spec.

5.2 Additional properties

While $value is the only required property for a token, a number of additional properties MAY be added:

5.3 Description

A plain text description explaining the token's purpose can be provided via the optional $description property. Tools MAY use the description in various ways.

For example:

The value of the $description property MUST be a plain JSON string, for example:

5.4 Type

Design tokens always have an unambiguous type, so that tools can reliably interpret their value.

A token's type can be specified by the optional $type property. If the $type property is not set on a token, then the token's type MUST be determined as follows:

Tools MUST NOT attempt to guess the type of a token by inspecting the contents of its value.

The $type property can be set on different levels:

The $type property MUST be a plain JSON string, whose value is one of the values specified in this specification's respective type definitions. The value of $type is case-sensitive.

For example:

5.5 Extensions

The optional $extensions property is an object where tools MAY add proprietary, user-, team- or vendor-specific data to a design token. When doing so, each tool MUST use a vendor-specific key whose value MAY be any valid JSON data.

In order to maintain interoperability between tools that support this format, teams and tools SHOULD restrict their usage of extension data to optional meta-data that is not crucial to understanding that token's value.

Tool vendors are encouraged to publicly share specifications of their extension data wherever possible. That way other tools can add support for them without needing to reverse engineer the extension data. Popular extensions could also be incorporated as standardized features in future revisions of this specification.

Editor's note: Extensions section

The extensions section is not limited to vendors. All token users can add additional data in this section for their own purposes.

5.6 More token properties TBC

6. Groups

A file MAY contain many tokens and they MAY be nested arbitrarily in groups like so:

The names of the groups leading to a given token (including that token's name) are that token's path, which is a computed property. It is not specified in the file, but parsers that conform to this spec MUST be able to expose the path of a token. The above example, therefore, defines 4 design tokens with the following properties:

Because groupings are arbitrary, tools MUST NOT use them to infer the type or purpose of design tokens.

Groups items (i.e. the tokens and/or nested groups) are unordered. In other words, there is no implicit order between items within a group. Therefore, tools that parse or write design token files are not required to preserve the source order of items in a group.

The names of items in a group are case sensitive. As per the guidance in the design token chapter, tools MAY display a warning to users when groups contain items whose names differ only in case and could therefore lead to naming clashes when exported.

Editor's note: Naming practices

The format editors acknowledge existing best-practices for token naming, but place no direct constraints on naming via the specification.

6.1 Additional group properties

Editor's note: Group properties vs. nested group and token names

To prevent collisions with token names, token properties are prefixed with a dollar sign ($). Using this prefix eliminates the need for a reserved words list and helps future-proof the spec.

Group keys without a dollar sign ($) prefix denote:

  • A token name: distinguishable by containing a $value property

    {
      "Group of tokens": {
        "$description": "This is an example of a group containing a single token",
        "Token name": {
          "$value": "#000000"
        }
      }
    }
    
  • A nested group name: distinguishable by not having a $value property

    {
      "Group of tokens": {
        "$description": "This is an example of a group containing a nested group",
        "Subgroup of tokens": {
          "Token 1 name": {
            "$value": "#aabbcc"
          },
          "Token 2 name": {
            "$value": "#ddeeff"
          }
        }
      }
    }
    

6.1.1 Description

Groups MAY include an optional $description property, whose value MUST be a plain JSON string. Its purpose is to describe the group itself.

For example:

Suggested ways tools MAY use this property are:

  • A style guide generator could render a section for each group and use the description as an introductory paragraph
  • A GUI tool that lets users browse or select tokens could display this info alongside the corresponding group or as a tooltip
  • Translation tools could output this as a source code comment

6.1.2 Type

Groups MAY include an optional $type property so a type property does not need to be manually added to every token. See supported "Types" for more information.

If a group has a $type property it acts as a default type for any tokens within the group, including ones in nested groups, that do not explicitly declare a type via their own $type property. For the full set of rules by which a design token's type is determined, please refer to the design token type property chapter.

For example:

6.1.3 Extensions

Groups MAY include an optional $extensions property where tools MAY add proprietary, user-, team- or vendor-specific data to a group. When doing so, each tool MUST use a vendor-specific key whose value MAY be any valid JSON data.

Group extensions follow the same rules as design token extensions, the only difference is that they relate to the group itself.

  • The keys SHOULD be chosen such that they avoid the likelihood of a naming clash with another vendor's data. The reverse domain name notation is recommended for this purpose.
  • Tools that process design token files MUST preserve any extension data they do not themselves understand. For example, if a group contains extension data from tool A and the file containing that data is opened by tool B, then tool B MUST include the original tool A extension data whenever it saves a new design token file containing that group.

Note that, since a group's $extensions only relate to that group, they do not have any effect on nested groups or tokens.

6.2 Use-cases

6.2.1 File authoring & organization

Groups let token file authors better organize their token files. Related tokens can be nested into groups to align with the team's naming conventions and/or mental model. When manually authoring files, using groups is also less verbose than a flat list of tokens with repeating prefixes.

For example:

...is likely to be more convenient to type and, arguably, easier to read, than:

6.2.2 GUI tools

Tools that let users pick or edit tokens via a GUI MAY use the grouping structure to display a suitable form of progressive disclosure, such as a collapsible tree view.

Figure 1 Progressive disclosure groups

6.2.3 Translation tools

Token names are not guaranteed to be unique within the same file. The same name can be used in different groups. Also, translation tools MAY need to export design tokens in a uniquely identifiable way, such as variables in code. Translation tools SHOULD therefore use design tokens' paths as these are unique within a file.

For example, a translation tool like Style Dictionary might use the following design token file:

...and output it as Sass variables like so by concatenating the path to create variable names:

7. Aliases / references

Instead of having explicit values, tokens can reference the value of another token. To put it another way, a token can be an alias for another token. This spec considers the terms "alias" and "reference" to be synonyms and uses them interchangeably.

Aliases are useful for:

For a design token to reference another, its value MUST be a string containing the period-separated (.) path to the token it's referencing enclosed in curly brackets.

For example:

When a tool needs the actual value of a token it MUST resolve the reference - i.e. lookup the token being referenced and fetch its value. In the above example, the "alias name" token's value would resolve to 1234 because it references the token whose path is {group name.token name} which has the value 1234.

Tools SHOULD preserve references and therefore only resolve them whenever the actual value needs to be retrieved. For instance, in a design tool, changes to the value of a token being referenced by aliases SHOULD be reflected wherever those aliases are being used.

Aliases MAY reference other aliases. In this case, tools MUST follow each reference until they find a token with an explicit value. Circular references are not allowed. If a design token file contains circular references, then the value of all tokens in that chain is unknown and an appropriate error or warning message SHOULD be displayed to the user.

Editor's note: JSON Pointer syntax

The format editors are currently researching JSON Pointer syntax to inform the exact syntax for aliases in tokens. https://datatracker.ietf.org/doc/html/rfc6901#section-5

8. Types

Many tools need to know what kind of value a given token represents to process it sensibly. Translation tools MAY need to convert or format tokens differently depending on their type. Design tools MAY present the user with different kinds of input when editing tokens of a certain type (such as color picker, slider, text input, etc.). Style guide generators MAY use different kinds of previews for different types of tokens.

This spec defines a number of design-focused types and every design token MUST use one of these types. Furthermore, that token's value MUST then follow rules and syntax for the chosen type as defined by this spec.

A token's type can be set directly by giving it a $type property specifying the chosen type. Alternatively, it can inherit a type from one of its parent groups, or be an alias of a token that has the desired type.

If no explicit type has been set for a token, tools MUST consider the token invalid and not attempt to infer any other type from the value.

If an explicit type is set, but the value does not match the expected syntax then that token is invalid and an appropriate error SHOULD be displayed to the user. To put it another way, the $type property is a declaration of what kind of values are permissible for the token. (This is similar to typing in programming languages like Java or TypeScript, where a value not compatible with the declared type causes a compilation error).

8.1 Color

Represents a 24bit RGB or 24+8bit RGBA color in the sRGB color space. The $type property MUST be set to the string color. The value MUST be a string containing a hex triplet/quartet including the preceding # character. To support other color spaces, such as HSL, translation tools SHOULD convert color tokens to the equivalent value as needed.

For example, initially the color tokens MAY be defined as such:

Then, the output from a tool's conversion to HSL(A) MAY look something like:

8.2 Dimension

Represents an amount of distance in a single dimension in the UI, such as a position, width, height, radius, or thickness. The $type property MUST be set to the string dimension. The value MUST be an object containing a numeric value (integer or floating-point) and unit of measurement ("px" or "rem").

Key Type Required Description
value number Y An integer or floating-point value representing the numeric value.
unit string Y Unit of distance. Supported values: "px", "rem".

For example:

8.2.1 Validation

  • $value.unit may only be "px" or "rem".
    • px: Represents an idealized pixel on the viewport. The equivalent in Android is dp and iOS is pt. Translation tools SHOULD therefore convert to these or other equivalent units as needed.
    • rem: Represents a multiple of the system's default font size (which MAY be configurable by the user). 1rem is 100% of the default font size. The equivalent of 1rem on Android is 16sp. Not all platforms have an equivalent to rem, so translation tools MAY need to do a lossy conversion to a fixed px size by assuming a default font size (usually 16px) for such platforms.
  • $value.unit is still required even if $value.value is 0.

8.3 Font family

A naive approach like the one below may be appropriate for the first stage of the specification, but this could be more complicated than it seems due to platform/OS/browser restrictions.

Represents a font name or an array of font names (ordered from most to least preferred). The $type property MUST be set to the string fontFamily. The value MUST either be a string value containing a single font name or an array of strings, each being a single font name.

For example:

8.4 Font weight

Represents a font weight. The $type property MUST be set to the string fontWeight. The value must either be a number value in the range [1, 1000] or one of the pre-defined string values defined in the table below.

Lower numbers represent lighter weights, and higher numbers represent thicker weights, as per the OpenType wght tag specification. The pre-defined string values are aliases for specific numeric values. For example 100, "thin" and "hairline" are all the exact same value.

numeric value string value aliases
100 thin, hairline
200 extra-light, ultra-light
300 light
400 normal, regular, book
500 medium
600 semi-bold, demi-bold
700 bold
800 extra-bold, ultra-bold
900 black, heavy
950 extra-black, ultra-black

Number values outside of the [1, 1000] range and any other string values, including ones that differ only in case, are invalid and MUST be rejected by tools.

Example:

8.5 Duration

Represents the length of time in milliseconds an animation or animation cycle takes to complete, such as 200 milliseconds. The $type property MUST be set to the string duration. The value MUST be an object containing a numeric value (either integer or floating-point) and a unit of milliseconds ("ms") or seconds ("s"). A millisecond is a unit of time equal to one thousandth of a second.

Key Type Required Description
value number Y An integer or floating-point value representing the numeric value.
unit string Y Unit of time. Supported values: "ms" (millisecond), "s"(second).

For example:

8.5.1 Validation

  • $value.unit may only be "ms" or "s"

8.6 Cubic Bézier

Represents how the value of an animated property progresses towards completion over the duration of an animation, effectively creating visual effects such as acceleration, deceleration, and bounce. The $type property MUST be set to the string cubicBezier. The value MUST be an array containing four numbers. These numbers represent two points (P1, P2) with one x coordinate and one y coordinate each [P1x, P1y, P2x, P2y]. The y coordinates of P1 and P2 can be any real number in the range [-∞, ∞], but the x coordinates are restricted to the range [0, 1].

For example:

8.7 Number

Represents a number. Numbers can be positive, negative and have fractions. Example uses for number tokens are gradient stop positions or unitless line heights. The $type property MUST be set to the string number. The value MUST be a JSON number value.

8.8 Additional types

This section is non-normative.

Types still to be documented here are likely to include:

9. Composite types

The types defined in the previous chapters such as color and dimension all have singular values. For example, the value of a color token is one color. However, there are other aspects of UI designs that are a combination of multiple values. For instance, a shadow style is a combination of a color, X & Y offsets, a blur radius and a spread radius.

Every shadow style has the exact same parts (color, X & Y offsets, etc.), but their respective values will differ. Furthermore, each part's value (which is also known as a "sub-value") is always of the same type. A shadow's color must always be a color value, its X offset must always be a dimension value, and so on. Shadow styles are therefore combinations of values that follow a pre-defined structure. In other words, shadow styles are themselves a type. Types like this are called composite types.

Specifically, a composite type has the following characteristics:

A design token whose type happens to be a composite type is sometimes also called a composite (design) token. Besides their type, there is nothing special about composite tokens. They can have all the other additional properties like $description or $extensions. They can also be referenced by other design tokens.

9.1 Groups versus composite tokens

At first glance, groups and composite tokens might look very similar. However, they are intended to solve different problems and therefore have some important differences:

9.2 Stroke style

Represents the style applied to lines or borders. The $type property MUST be set to the string strokeStyle. The value MUST be either:

Issue 98: Stroke style type feedback Composite Type Feedback
Is the current specification for stroke styles fit for purpose? Does it need more sub-values (e.g. equivalents to SVG's stroke-linejoin, stroke-miterlimit and stroke-dashoffset attributes)?

9.2.1 String value

String stroke style values MUST be set to one of the following, pre-defined values:

  • solid
  • dashed
  • dotted
  • double
  • groove
  • ridge
  • outset
  • inset

These values have the same meaning as the equivalent "line style" values in CSS. As per the CSS spec, their exact rendering is therefore implementation specific. For example, the length of dashes and gaps in the dashed style may vary between different tools.

9.2.2 Object value

Object stroke style values MUST have the following properties:

  • dashArray: An array of dimension values and/or references to dimension tokens, which specify the lengths of alternating dashes and gaps. If an odd number of values is provided, then the list of values is repeated to yield an even number of values.
  • lineCap: One of the following pre-defined string values: "round", "butt" or "square". These values have the same meaning as those of the stroke-linecap attribute in SVG.

9.2.3 Fallbacks

The string and object values are mutually exclusive means of expressing stroke styles. For example, some of the string values like inset or groove cannot be expressed in terms of a dashArray and lineCap as they require some implementation-specific means of lightening or darkening the color for portions of a border or outline. Conversely, a precisely defined combination of dashArray and lineCap sub-values is not guaranteed to produce the same visual result as the dashed or dotted keywords as they are implementation-specific.

Furthermore, some tools and platforms may not support the full range of stroke styles that design tokens of this type can represent. When displaying or exporting a strokeStyle token whose value they don't natively support, they should therefore fallback to the closest approximation that they do support.

The specifics of how a "closest approximation" is chosen are implementation-specific. However, the following examples illustrate what fallbacks tools MAY use in some scenarios.

9.3 Border

Represents a border style. The $type property MUST be set to the string border. The value MUST be an object with the following properties:

Issue 99: Border type feedback Composite Type Feedback
Is the current specification for borders fit for purpose? Does it need more sub-values to account for features like outset, border images, multiple borders, etc. that some platforms and design tools have?

9.4 Transition

Represents a animated transition between two states. The $type property MUST be set to the string transition. The value MUST be an object with the following properties:

Issue 103: Transition type feedback Composite Type Feedback
Is the current specification for transitions fit for purpose? Are these transitions parameters by themselves useful considering that they don't let you specify what aspect of a UI is being transitioned and what the start and end states are?

9.5 Shadow

Represents a shadow style. The $type property MUST be set to the string shadow. The value MUST contain a single object or an array of objects with the following properties:

Issue 100: Shadow type feedback Composite Type Feedback
Is the current specification for shadows fit for purpose? Does it need to support multiple shadows, as some tools and platforms do?

9.6 Gradient

Represents a color gradient. The $type property MUST be set to the string gradient. The value MUST be an array of objects representing gradient stops that have the following structure:

If there are no stops at the very beginning or end of the gradient axis (i.e. with position 0 or 1, respectively), then the color from the stop closest to each end should be extended to that end of the axis.

Issue 101: Gradient type feedback Color Type Enhancements
Is the current specification for gradients fit for purpose? Does it need to also specify the type of gradient (.e.g linear, radial, conical, etc.)?

9.7 Typography

Represents a typographic style. The $type property MUST be set to the string typography. The value MUST be an object with the following properties:

Issue 102: Typography type feedback Typography Type Enhancements

Is the current specification for typography styles fit for purpose? Should the lineHeight sub-value use a number value, dimension or a new lineHeight type?

A. Issue summary

B. References

B.1 Normative references

[RFC2119]
Key words for use in RFCs to Indicate Requirement Levels. S. Bradner. IETF. March 1997. Best Current Practice. URL: https://www.rfc-editor.org/rfc/rfc2119
[RFC8174]
Ambiguity of Uppercase vs Lowercase in RFC 2119 Key Words. B. Leiba. IETF. May 2017. Best Current Practice. URL: https://www.rfc-editor.org/rfc/rfc8174