Bridge

Defined in: ai.matey.core/src/bridge.ts:48

Bridge connects frontend and backend adapters.

Type Parameters

TFrontend

TFrontend extends FrontendAdapter = FrontendAdapter

Frontend adapter type

Implements

Bridge<TFrontend>

Constructors

Constructor

new Bridge<TFrontend>(frontend, backend, config): Bridge<TFrontend>

Defined in: ai.matey.core/src/bridge.ts:76

Create a new Bridge instance.

Parameters

frontend

TFrontend

Frontend adapter

backend

BackendAdapter

Backend adapter

config

Partial<BridgeConfig> = {}

Bridge configuration

Returns

Bridge<TFrontend>

Properties

backend

readonly backend: BackendAdapter

Defined in: ai.matey.core/src/bridge.ts:52

Backend adapter or router.

Implementation of

Bridge.backend

config

readonly config: BridgeConfig

Defined in: ai.matey.core/src/bridge.ts:53

Bridge configuration.

Implementation of

Bridge.config

frontend

readonly frontend: TFrontend

Defined in: ai.matey.core/src/bridge.ts:51

Frontend adapter for this bridge.

Implementation of

Bridge.frontend

generateObject()

generateObject: <T>(options) => Promise<GenerateObjectResult<T>>

Defined in: ai.matey.core/src/bridge.ts:673

Generate a structured object matching a Zod schema using an LLM.

This method uses tool calling to extract structured data from the LLM response, validates it against the provided schema, and returns the typed object.

Type Parameters

T

T = any

Parameters

options

GenerateObjectOptions

Configuration for object generation

Returns

Promise<GenerateObjectResult<T>>

Promise resolving to the generated and validated object

Example

const UserSchema = z.object({
  name: z.string(),
  age: z.number(),
  email: z.string().email(),
});

const result = await bridge.generateObject({
  schema: UserSchema,
  prompt: 'Generate a user profile for Alice, age 30',
  model: 'gpt-4',
});

console.log(result.object); // { name: 'Alice', age: 30, email: '...' }

streamObject()

streamObject: <T>(options) => AsyncGenerator<Partial<T>, T>

Defined in: ai.matey.core/src/bridge.ts:705

Stream a structured object matching a Zod schema using an LLM.

This method uses streaming tool calling to incrementally build up a structured object, yielding partial results as they become available.

Type Parameters

T

T = any

Parameters

options

StreamObjectOptions

Configuration for streaming object generation

Returns

AsyncGenerator<Partial<T>, T>

Async generator yielding partial objects and returning the final validated object

Example

const ArticleSchema = z.object({
  title: z.string(),
  content: z.string(),
  tags: z.array(z.string()),
});

const stream = bridge.streamObject({
  schema: ArticleSchema,
  prompt: 'Write a blog post about TypeScript',
  onPartial: (partial) => {
    console.log('Partial:', partial);
  },
});

for await (const partial of stream) {
  console.log('Progress:', partial);
}

Methods

chat()

chat(request, options?): Promise<InferFrontendResponse<TFrontend>>

Defined in: ai.matey.core/src/bridge.ts:98

Execute a non-streaming chat completion request.

Parameters

request

InferFrontendRequest<TFrontend>

options?

RequestOptions

Returns

Promise<InferFrontendResponse<TFrontend>>

Implementation of

Bridge.chat

chatStream()

chatStream(request, options?): AsyncGenerator<InferFrontendStreamChunk<TFrontend>, void, undefined>

Defined in: ai.matey.core/src/bridge.ts:216

Execute a streaming chat completion request.

Parameters

request

InferFrontendRequest<TFrontend>

options?

RequestOptions

Returns

AsyncGenerator<InferFrontendStreamChunk<TFrontend>, void, undefined>

Implementation of

Bridge.chatStream

checkHealth()

checkHealth(): Promise<boolean>

Defined in: ai.matey.core/src/bridge.ts:550

Check health of the backend.

Returns

Promise<boolean>

true if backend is healthy, false otherwise

clearMiddleware()

clearMiddleware(): Bridge<TFrontend>

Defined in: ai.matey.core/src/bridge.ts:413

Clear all middleware from the stack.

Returns

Bridge<TFrontend>

Implementation of

Bridge.clearMiddleware

clone()

clone(config): Bridge<TFrontend>

Defined in: ai.matey.core/src/bridge.ts:569

Clone bridge with new configuration.

Parameters

config

Partial<BridgeConfig>

Returns

Bridge<TFrontend>

Implementation of

Bridge.clone

dispose()

dispose(): void

Defined in: ai.matey.core/src/bridge.ts:579

Clean up resources.

Returns

void

Implementation of

Bridge.dispose

getConfig()

getConfig(): Readonly<BridgeConfig>

Defined in: ai.matey.core/src/bridge.ts:562

Get a copy of the bridge configuration.

Returns

Readonly<BridgeConfig>

Readonly copy of the bridge configuration

getMiddleware()

getMiddleware(): readonly Middleware[]

Defined in: ai.matey.core/src/bridge.ts:421

Get all middleware in the stack.

Returns

readonly Middleware[]

Implementation of

Bridge.getMiddleware

getRouter()

getRouter(): null

Defined in: ai.matey.core/src/bridge.ts:541

Get router instance (returns null for basic Bridge).

Returns

null

Implementation of

Bridge.getRouter

getStats()

getStats(): BridgeStats

Defined in: ai.matey.core/src/bridge.ts:488

Get runtime statistics for this bridge.

Returns

BridgeStats

Bridge statistics including request counts, latencies, and error breakdown

Implementation of

Bridge.getStats

hasModel()

hasModel(modelId): Promise<boolean>

Defined in: ai.matey.core/src/bridge.ts:347

Check if a specific model is available from the backend.

Parameters

modelId

string

Model identifier to check

Returns

Promise<boolean>

true if model is available, false otherwise

listModels()

listModels(options?): Promise<ListModelsResult | null>

Defined in: ai.matey.core/src/bridge.ts:333

List available models from the backend.

This delegates directly to the backend adapter’s listModels() method. Useful for discovering available models before making requests.

Parameters

options?

ListModelsOptions

Options for listing models (filtering, cache control)

Returns

Promise<ListModelsResult | null>

List of available models, or null if backend doesn’t support listing

off()

off(event, listener): Bridge<TFrontend>

Defined in: ai.matey.core/src/bridge.ts:453

Remove an event listener.

Parameters

event

Event type

BridgeEventType | "*"

listener

BridgeEventListener

Callback function to remove

Returns

Bridge<TFrontend>

Implementation of

Bridge.off

on()

on(event, listener): Bridge<TFrontend>

Defined in: ai.matey.core/src/bridge.ts:438

Note: Event emission is not yet implemented. Listeners are stored for future use when event emission is added.

Parameters

event

Event type to listen for, or ‘*’ for all events

BridgeEventType | "*"

listener

BridgeEventListener

Callback function

Returns

Bridge<TFrontend>

Implementation of

Bridge.on

once()

once(event, listener): Bridge<TFrontend>

Defined in: ai.matey.core/src/bridge.ts:471

Note: Event emission is not yet implemented. Listeners are stored for future use when event emission is added.

Parameters

event

BridgeEventType

Event type to listen for

listener

BridgeEventListener

Callback function

Returns

Bridge<TFrontend>

Implementation of

Bridge.once

removeMiddleware()

removeMiddleware(middleware): Bridge<TFrontend>

Defined in: ai.matey.core/src/bridge.ts:405

Remove middleware from the stack.

Parameters

middleware

Middleware

Middleware to remove

Returns

Bridge<TFrontend>

This bridge for chaining

Implementation of

Bridge.removeMiddleware

resetStats()

resetStats(): void

Defined in: ai.matey.core/src/bridge.ts:524

Reset all statistics to zero.

Returns

void

Implementation of

Bridge.resetStats

use()

use(middleware): Bridge<TFrontend>

Defined in: ai.matey.core/src/bridge.ts:394

Add middleware to the bridge’s middleware stack.

Parameters

middleware

Middleware

Returns

Bridge<TFrontend>

Implementation of

Bridge.use

validateModel()

validateModel(modelId): Promise<void>

Defined in: ai.matey.core/src/bridge.ts:365

Validate that a model is available (optional safety check).

Note: This is an optional validation - the system doesn’t automatically validate models since cross-provider translation is supported.

Parameters

modelId

string

Model identifier to validate

Returns

Promise<void>

Throws

If model is not available