Class Skill

Skill schema defines the structure for AI assistant skills. Skills contain instructions, tools, and artifacts that guide the AI's behavior. Skills may use tools to create and read artifacts, which are managed by the assistant.

Follows the Agent Skills spec (https://agentskills.io/specification) as closely as it can, so a skill means the same here as in any other harness. Add a field the spec lacks only with a strong reason, and say what it is. Credentials and MCP servers do not belong here: a server is a connection, which the space owns and each conversation can turn off. Nor do hooks: a plugin contributes those beside its skills (@dxos/compute/Hook).

The registry key and version are stored in the object meta — access them via Obj.getMeta(skill).key and Obj.getMeta(skill).version.

Hierarchy

{
    agentCanEnable?: boolean;
    description?: string;
    instructions: {
        inputs?: readonly {
            default?: any;
            kind: "value" | "operation";
            name: string;
            operation?: string;
        }[];
        source: Ref<Text & OfKind<Object> & Unknown>;
    };
    name: string;
    tools: readonly (string & Brand<"ToolId">)[];
} & OfKind<Object, this>
- Skill

Index

Constructors

constructor

Properties

~@dxos/echo/Kind agentCanEnable? description? id instructions name tools ~@dxos/echo/Kind ~@dxos/echo/SchemaKind ~@dxos/echo/Type.Instance? ~@dxos/echo/Type.StaticSchema fields id jsonSchema name?

Constructors

constructor

new Skill(_: never): Skill.Skill
Parameters
- _: never
Returns Skill.Skill
Inherited from Type.makeObject<Skill>(DXN.make('org.dxos.type.skill', '0.2.0'))( Schema.Struct({ /** * Human-readable name of the skill. */ name: Schema.String.annotate({ description: 'Human-readable name of the skill', }), /** * Description of the skill's purpose and functionality. */ description: Schema.optional(Format.Text).annotate({ description: "Description of the skill's purpose and functionality", }), /** * Instructions that guide the AI assistant's behavior and responses. * These are system prompts or guidelines that the AI should follow. */ instructions: Template.Template.annotate({ description: "Instructions that guide the AI assistant's behavior and responses", }), /** * Array of tools that the AI assistant can use when this skill is active. */ tools: Schema.Array(ToolId).annotate({ description: 'Array of tools that the AI assistant can use when this skill is active', }), /** * Whether an agent is allowed to auto-enable this skill in a conversation. */ agentCanEnable: Schema.optional(Schema.Boolean).annotate({ description: 'Whether an agent is allowed to auto-enable this skill in a conversation.', }), }).pipe( Annotation.LabelAnnotation.set(['name']), Annotation.IconAnnotation.set({ icon: 'ph--student--regular', hue: 'amber' }), Annotation.UserType.set(), ), ).constructor
- Defined in packages/core/echo/echo/dist/types/src/Type.d.ts:72

Properties

`Readonly`~@dxos/echo/Kind

"~@dxos/echo/Kind": Object

Inherited from Type.makeObject<Skill>(DXN.make('org.dxos.type.skill', '0.2.0'))( Schema.Struct({ /** * Human-readable name of the skill. */ name: Schema.String.annotate({ description: 'Human-readable name of the skill', }), /** * Description of the skill's purpose and functionality. */ description: Schema.optional(Format.Text).annotate({ description: "Description of the skill's purpose and functionality", }), /** * Instructions that guide the AI assistant's behavior and responses. * These are system prompts or guidelines that the AI should follow. */ instructions: Template.Template.annotate({ description: "Instructions that guide the AI assistant's behavior and responses", }), /** * Array of tools that the AI assistant can use when this skill is active. */ tools: Schema.Array(ToolId).annotate({ description: 'Array of tools that the AI assistant can use when this skill is active', }), /** * Whether an agent is allowed to auto-enable this skill in a conversation. */ agentCanEnable: Schema.optional(Schema.Boolean).annotate({ description: 'Whether an agent is allowed to auto-enable this skill in a conversation.', }), }).pipe( Annotation.LabelAnnotation.set(['name']), Annotation.IconAnnotation.set({ icon: 'ph--student--regular', hue: 'amber' }), Annotation.UserType.set(), ), ).~@dxos/echo/Kind

Defined in packages/core/echo/echo/dist/types/src/Entity.d.ts:20

`Optional` `Readonly`agentCanEnable

agentCanEnable?: boolean = ...

Whether an agent is allowed to auto-enable this skill in a conversation.

Defined in packages/core/compute/compute/src/types/Skill.ts:70

`Optional` `Readonly`description

description?: string = ...

Description of the skill's purpose and functionality.

Defined in packages/core/compute/compute/src/types/Skill.ts:48

`Readonly`id

id: string

Defined in packages/core/echo/echo/dist/types/src/Entity.d.ts:21

`Readonly`instructions

instructions: {
    inputs?: readonly {
        default?: any;
        kind: "value" | "operation";
        name: string;
        operation?: string;
    }[];
    source: Ref<Text & OfKind<Object> & Unknown>;
} = ...

Instructions that guide the AI assistant's behavior and responses. These are system prompts or guidelines that the AI should follow.

Type declaration

Optional Readonlyinputs?: readonly {
    default?: any;
    kind: "value" | "operation";
    name: string;
    operation?: string;
}[]
NOTE: We use an array rather than map so that updating variable names in the template doesn't disconnect existing inputs.
Readonlysource: Ref<Text & OfKind<Object> & Unknown>

Defined in packages/core/compute/compute/src/types/Skill.ts:56

`Readonly`name

name: string = ...

Human-readable name of the skill.

Defined in packages/core/compute/compute/src/types/Skill.ts:41

`Readonly`tools

tools: readonly (string & Brand<"ToolId">)[] = ...

Array of tools that the AI assistant can use when this skill is active.

Defined in packages/core/compute/compute/src/types/Skill.ts:63

`Static` `Readonly`~@dxos/echo/Kind

"~@dxos/echo/Kind": Type

Entity-kind brand of the type-entity value itself — always EntityKind.Type. The kind of instance the type describes lives on [SchemaKindId] (Object / Relation / Type). Lets Obj.isObject / Relation.isRelation reject type entities by a single [KindId] check.

Defined in packages/core/echo/echo/dist/types/src/Type.d.ts:23

`Static` `Readonly`~@dxos/echo/SchemaKind

"~@dxos/echo/SchemaKind": Object

Schema-kind brand (object).

Defined in packages/core/echo/echo/dist/types/src/Type.d.ts:57

`Static` `Optional` `Readonly`~@dxos/echo/Type.Instance

"~@dxos/echo/Type.Instance"?: Skill.Skill & OfKind<Object>

Defined in packages/core/echo/echo/dist/types/src/Type.d.ts:34

`Static` `Readonly`~@dxos/echo/Type.StaticSchema

"~@dxos/echo/Type.StaticSchema": Codec<any, any>

Source Effect Schema — used internally by Type.getSchema(self).

Defined in packages/core/echo/echo/dist/types/src/Type.d.ts:59

`Static` `Readonly`fields

fields: {}

The fields defined in the original struct schema. Allows accessing field definitions for introspection.

Defined in packages/core/echo/echo/dist/types/src/Type.d.ts:64

`Static` `Readonly`id

id: string

Object id. Like all ECHO entities, type entities always carry an id — stamped at construction for in-memory (static) declarations and assigned by the database once persisted. The id does NOT determine the entity's URI: static types resolve to their typename DXN, persisted types to echo:///<id> (see getTypeURIFromSpecifier).

Defined in packages/core/echo/echo/dist/types/src/Type.d.ts:31

`Static` `Readonly`jsonSchema

jsonSchema: JsonSchemaType

Defined in packages/core/echo/echo/dist/types/src/Type.d.ts:33

`Static` `Optional` `Readonly`name

name?: string

Defined in packages/core/echo/echo/dist/types/src/Type.d.ts:32

Class Skill

Hierarchy

Index

Constructors

Properties

Constructors

constructor

Parameters

Returns Skill.Skill

Properties

`Readonly`~@dxos/echo/Kind

`Optional` `Readonly`agentCanEnable

`Optional` `Readonly`description

`Readonly`id

`Readonly`instructions

Type declaration

`Optional` `Readonly`inputs?: readonly {
    default?: any;
    kind: "value" | "operation";
    name: string;
    operation?: string;
}[]

`Readonly`source: Ref<Text & OfKind<Object> & Unknown>

`Readonly`name

`Readonly`tools

`Static` `Readonly`~@dxos/echo/Kind

`Static` `Readonly`~@dxos/echo/SchemaKind

`Static` `Optional` `Readonly`~@dxos/echo/Type.Instance

`Static` `Readonly`~@dxos/echo/Type.StaticSchema

`Static` `Readonly`fields

`Static` `Readonly`id

`Static` `Readonly`jsonSchema

`Static` `Optional` `Readonly`name

Settings

On This Page

Class Skill

Hierarchy

Index

Constructors

Properties

Constructors

constructor

Parameters

Returns Skill.Skill

Properties

Readonly~@dxos/echo/Kind

Optional ReadonlyagentCanEnable

Optional Readonlydescription

Readonlyid

Readonlyinstructions

Type declaration

Optional Readonlyinputs?: readonly { default?: any; kind: "value" | "operation"; name: string; operation?: string;}[]

Readonlysource: Ref<Text & OfKind<Object> & Unknown>

Readonlyname

Readonlytools

Static Readonly~@dxos/echo/Kind

Static Readonly~@dxos/echo/SchemaKind

Static Optional Readonly~@dxos/echo/Type.Instance

Static Readonly~@dxos/echo/Type.StaticSchema

Static Readonlyfields

Static Readonlyid

Static ReadonlyjsonSchema

Static Optional Readonlyname

Settings

On This Page

`Readonly`~@dxos/echo/Kind

`Optional` `Readonly`agentCanEnable

`Optional` `Readonly`description

`Readonly`id

`Readonly`instructions

`Optional` `Readonly`inputs?: readonly {
default?: any;
kind: "value" | "operation";
name: string;
operation?: string;
}[]

`Readonly`source: Ref<Text & OfKind<Object> & Unknown>

`Readonly`name

`Readonly`tools

`Static` `Readonly`~@dxos/echo/Kind

`Static` `Readonly`~@dxos/echo/SchemaKind

`Static` `Optional` `Readonly`~@dxos/echo/Type.Instance

`Static` `Readonly`~@dxos/echo/Type.StaticSchema

`Static` `Readonly`fields

`Static` `Readonly`id

`Static` `Readonly`jsonSchema

`Static` `Optional` `Readonly`name