Class ActionEmbeddingService

Utility service for action embedding operations. Uses action data already loaded by AIEngine - no separate initialization needed. All methods are static - no instance state required.

Follows the same pattern as AgentEmbeddingService for consistency.

Example

// Generate embeddings during AIEngine initialization
const entries = await ActionEmbeddingService.GenerateActionEmbeddings(
  actions,
  (text) => aiEngine.EmbedTextLocal(text)
);

// Find similar actions
const matches = await ActionEmbeddingService.FindSimilarActions(
  vectorService,
  "Search the web for information",
  (text) => aiEngine.EmbedTextLocal(text),
  5,
  0.7
);

Index

Constructors

constructor

Methods

Constructors

constructor

new ActionEmbeddingService(): ActionEmbeddingService
Returns ActionEmbeddingService

Methods

`Static`FindRelatedActions

FindRelatedActions(
    vectorService: SimpleVectorService<ActionEmbeddingMetadata>,
    actionId: string,
    topK?: number,
    minSimilarity?: number,
): ActionMatchResult[]
Find actions similar to a specific action.
Parameters
- vectorService: SimpleVectorService<ActionEmbeddingMetadata>
  Vector service containing action embeddings
- actionId: string
  The ID of the action to find similar actions to
- topK: number = 5
  Maximum number of results to return (default: 5)
- minSimilarity: number = 0.5
  Minimum similarity score 0-1 (default: 0.5)
Returns ActionMatchResult[]
Array of matching actions (excludes the source action)

Throws
Error if action not found in embeddings
- Defined in AI/Engine/src/services/ActionEmbeddingService.ts:157

`Static`FindSimilarActions

FindSimilarActions(
    vectorService: SimpleVectorService<ActionEmbeddingMetadata>,
    taskDescription: string,
    embedFunction: (
        text: string,
    ) => Promise<{ model: MJAIModelEntityExtended; result: EmbedTextResult }>,
    topK?: number,
    minSimilarity?: number,
): Promise<ActionMatchResult[]>
Find actions similar to a given task description.
Parameters
- vectorService: SimpleVectorService<ActionEmbeddingMetadata>
  Vector service containing action embeddings
- taskDescription: string
  The task description to match against action capabilities
- embedFunction: (
  text: string,
  ) => Promise<{ model: MJAIModelEntityExtended; result: EmbedTextResult }>
  Function to generate embeddings (from AIEngine)
- topK: number = 5
  Maximum number of results to return (default: 5)
- minSimilarity: number = 0.5
  Minimum similarity score 0-1 (default: 0.5)
Returns Promise<ActionMatchResult[]>
Array of matching actions sorted by similarity score (highest first)

Throws
Error if embedding generation fails
- Defined in AI/Engine/src/services/ActionEmbeddingService.ts:94

`Static`GenerateActionEmbeddings

GenerateActionEmbeddings(
    actions: MJActionEntity[],
    embedFunction: (
        text: string,
    ) => Promise<{ model: MJAIModelEntityExtended; result: EmbedTextResult }>,
): Promise<VectorEntry<ActionEmbeddingMetadata>[]>
Generate embeddings for actions using provided embedding function. Called by AIEngine during AdditionalLoading.
Parameters
- actions: MJActionEntity[]
  Array of actions (already loaded by AIEngine)
- embedFunction: (
  text: string,
  ) => Promise<{ model: MJAIModelEntityExtended; result: EmbedTextResult }>
  Function to generate embeddings (from AIEngine)
Returns Promise<VectorEntry<ActionEmbeddingMetadata>[]>
Vector entries ready for loading into vector service
- Defined in AI/Engine/src/services/ActionEmbeddingService.ts:42

`Static`GetSimilarity

GetSimilarity(
    vectorService: SimpleVectorService<ActionEmbeddingMetadata>,
    actionId1: string,
    actionId2: string,
): number
Get the similarity score between two specific actions.
Parameters
- vectorService: SimpleVectorService<ActionEmbeddingMetadata>
  Vector service containing action embeddings
- actionId1: string
  First action ID
- actionId2: string
  Second action ID
Returns number
Similarity score (0-1)

Throws
Error if either action is not found
- Defined in AI/Engine/src/services/ActionEmbeddingService.ts:201

Class ActionEmbeddingService

Example

Index

Constructors

Methods

Constructors

constructor

Returns ActionEmbeddingService

Methods

`Static`FindRelatedActions

Parameters

Returns ActionMatchResult[]

Throws

`Static`FindSimilarActions

Parameters

Returns Promise<ActionMatchResult[]>

Throws

`Static`GenerateActionEmbeddings

Parameters

Returns Promise<VectorEntry<ActionEmbeddingMetadata>[]>

`Static`GetSimilarity

Parameters

Returns number

Throws

Settings

On This Page

Class ActionEmbeddingService

Example

Index

Constructors

Methods

Constructors

constructor

Returns ActionEmbeddingService

Methods

StaticFindRelatedActions

Parameters

Returns ActionMatchResult[]

Throws

StaticFindSimilarActions

Parameters

Returns Promise<ActionMatchResult[]>

Throws

StaticGenerateActionEmbeddings

Parameters

Returns Promise<VectorEntry<ActionEmbeddingMetadata>[]>

StaticGetSimilarity

Parameters

Returns number

Throws

Settings

On This Page

`Static`FindRelatedActions

`Static`FindSimilarActions

`Static`GenerateActionEmbeddings

`Static`GetSimilarity