@agentskit/adapters — Interfaces

#Interface: AnthropicConfig

Defined in: adapters/src/anthropic.ts:6

#Properties

#apiKey

apiKey: string

Defined in: adapters/src/anthropic.ts:7

#baseUrl?

optional baseUrl?: string

Defined in: adapters/src/anthropic.ts:9

#maxTokens?

optional maxTokens?: number

Defined in: adapters/src/anthropic.ts:10

#model

model: string

Defined in: adapters/src/anthropic.ts:8

#retry?

optional retry?: RetryOptions

Defined in: adapters/src/anthropic.ts:11

#Interface: ApplyCarbonOptions

Defined in: adapters/src/carbon.ts:67

#Properties

#fallback?

optional fallback?: number

Defined in: adapters/src/carbon.ts:77

Fallback gCO2eq per 1k tokens when no table entry is found.

#keyFor?

optional keyFor?: (candidate) => `$\{string\}:$\{string\}` | undefined

Defined in: adapters/src/carbon.ts:75

How to derive the lookup key for a candidate. Defaults to $\{provider\}:$\{region\} where provider is the candidate id before the first - and region is candidate.region.

#Parameters

candidate

RouterCandidate

#Returns

`$\{string\}:$\{string\}` | undefined

#table?

optional table?: CarbonTable

Defined in: adapters/src/carbon.ts:69

Override the default table.

#Interface: AzureOpenAIConfig

Defined in: adapters/src/azure-openai.ts:5

#Properties

#apiKey

apiKey: string

Defined in: adapters/src/azure-openai.ts:6

#apiVersion?

optional apiVersion?: string

Defined in: adapters/src/azure-openai.ts:12

Azure REST api-version. Defaults to 2024-10-21.

#deployment

deployment: string

Defined in: adapters/src/azure-openai.ts:10

Deployment name (NOT the underlying model name — Azure routes by deployment).

#endpoint

endpoint: string

Defined in: adapters/src/azure-openai.ts:8

Resource endpoint, e.g. https://my-resource.openai.azure.com.

#includeUsage?

optional includeUsage?: boolean

Defined in: adapters/src/azure-openai.ts:15

Surface usage by setting stream_options.include_usage. Defaults to true.

#retry?

optional retry?: RetryOptions

Defined in: adapters/src/azure-openai.ts:13

#Interface: BailConfig

Defined in: adapters/src/bail.ts:4

#Extends

OpenAICompatibleConfig

#Properties

#apiKey

apiKey: string

Defined in: adapters/src/openai.ts:6

#Inherited from

OpenAICompatibleConfig.apiKey

#baseUrl?

optional baseUrl?: string

Defined in: adapters/src/openai.ts:8

#Inherited from

OpenAICompatibleConfig.baseUrl

#capabilities?

optional capabilities?: Partial<AdapterCapabilities>

Defined in: adapters/src/openai.ts:19

Explicit capability facts for OpenAI-compatible providers with custom models.

#Inherited from

OpenAICompatibleConfig.capabilities

#includeUsage?

optional includeUsage?: boolean

Defined in: adapters/src/openai.ts:17

Ask the provider to include token usage in the final stream chunk via stream_options: \{ include_usage: true \}. Off by default because some OpenAI-compatible providers (OpenRouter proxies to a long tail of backends) reject unknown params with a 4xx and break the whole stream. Turn this on for vanilla api.openai.com.

#Inherited from

OpenAICompatibleConfig.includeUsage

#model

model: string

Defined in: adapters/src/openai.ts:7

#Inherited from

OpenAICompatibleConfig.model

#retry?

optional retry?: RetryOptions

Defined in: adapters/src/openai.ts:9

#Inherited from

OpenAICompatibleConfig.retry

#Interface: BedrockConfig

Defined in: adapters/src/bedrock.ts:6

#Properties

#client?

optional client?: BedrockRuntimeClientLike

Defined in: adapters/src/bedrock.ts:18

Override the SDK client. Mostly for tests; production code should let the adapter create the client from region + the SDK's default credential chain.

#maxTokens?

optional maxTokens?: number

Defined in: adapters/src/bedrock.ts:12

Override max_tokens (Anthropic on Bedrock requires it). Defaults to 4096.

#model

model: string

Defined in: adapters/src/bedrock.ts:8

Bedrock model id, e.g. anthropic.claude-3-5-sonnet-20241022-v2:0.

#region?

optional region?: string

Defined in: adapters/src/bedrock.ts:10

AWS region (e.g. us-east-1). Falls back to the SDK's default credential chain.

#Interface: BedrockRuntimeClientLike

Defined in: adapters/src/bedrock.ts:28

Minimal structural type for BedrockRuntimeClient so we don't take a hard dep on @aws-sdk/client-bedrock-runtime.

Second options argument is optional so existing one-argument injected clients remain structurally valid.

#Methods

#send()

send(command, options?): Promise<{ body?: AsyncIterable<{ chunk?: { bytes?: Uint8Array<ArrayBufferLike>; }; }, any, any>; }>

Defined in: adapters/src/bedrock.ts:29

#Parameters

command

input

BedrockInvokeInput

options?

abortSignal?

AbortSignal

#Returns

Promise<{ body?: AsyncIterable<{ chunk?: { bytes?: Uint8Array<ArrayBufferLike>; }; }, any, any>; }>

#Interface: CerebrasConfig

Defined in: adapters/src/cerebras.ts:4

#Extends

OpenAICompatibleConfig

#Properties

#apiKey

apiKey: string

Defined in: adapters/src/openai.ts:6

#Inherited from

OpenAICompatibleConfig.apiKey

#baseUrl?

optional baseUrl?: string

Defined in: adapters/src/openai.ts:8

#Inherited from

OpenAICompatibleConfig.baseUrl

#capabilities?

optional capabilities?: Partial<AdapterCapabilities>

Defined in: adapters/src/openai.ts:19

Explicit capability facts for OpenAI-compatible providers with custom models.

#Inherited from

OpenAICompatibleConfig.capabilities

#includeUsage?

optional includeUsage?: boolean

Defined in: adapters/src/openai.ts:17

#Inherited from

OpenAICompatibleConfig.includeUsage

#model

model: string

Defined in: adapters/src/openai.ts:7

#Inherited from

OpenAICompatibleConfig.model

#retry?

optional retry?: RetryOptions

Defined in: adapters/src/openai.ts:9

#Inherited from

OpenAICompatibleConfig.retry

#Interface: CohereConfig

Defined in: adapters/src/cohere.ts:4

#Extends

OpenAICompatibleConfig

#Properties

#apiKey

apiKey: string

Defined in: adapters/src/openai.ts:6

#Inherited from

OpenAICompatibleConfig.apiKey

#baseUrl?

optional baseUrl?: string

Defined in: adapters/src/openai.ts:8

#Inherited from

OpenAICompatibleConfig.baseUrl

#capabilities?

optional capabilities?: Partial<AdapterCapabilities>

Defined in: adapters/src/openai.ts:19

Explicit capability facts for OpenAI-compatible providers with custom models.

#Inherited from

OpenAICompatibleConfig.capabilities

#includeUsage?

optional includeUsage?: boolean

Defined in: adapters/src/openai.ts:17

#Inherited from

OpenAICompatibleConfig.includeUsage

#model

model: string

Defined in: adapters/src/openai.ts:7

#Inherited from

OpenAICompatibleConfig.model

#retry?

optional retry?: RetryOptions

Defined in: adapters/src/openai.ts:9

#Inherited from

OpenAICompatibleConfig.retry

#Interface: CreateAdapterConfig

Defined in: adapters/src/types.ts:8

Optional second AbortSignal keeps one-argument callbacks assignable (extra parameters are optional). createAdapter races send against abort and will not start work after pre-abort.

#Properties

#abort?

optional abort?: () => void

Defined in: adapters/src/types.ts:17

#Returns

void

#parse

parse: (stream, response?) => AsyncIterableIterator<StreamChunk>

Defined in: adapters/src/types.ts:13

#Parameters

stream

ReadableStream

response?

Response

#Returns

AsyncIterableIterator<StreamChunk>

#send

send: (request, signal?) => Promise<ReadableStream<any> | Response>

Defined in: adapters/src/types.ts:9

#Parameters

request

AdapterRequest

signal?

AbortSignal

#Returns

Promise<ReadableStream<any> | Response>

#Interface: CredentialRefreshable

Defined in: adapters/src/credential-rotation.ts:39

#Properties

#refreshCredentials

refreshCredentials: (next) => Promise<void>

Defined in: adapters/src/credential-rotation.ts:46

Opt-in adapters that support credential rotation expose this method. Stock AgentsKit adapters do not implement it unless documented otherwise. Calling it replaces the in-memory secret. The next request uses the new value; in-flight requests are unaffected.

#Parameters

string

#Returns

Promise<void>

#Interface: DeepSeekConfig

Defined in: adapters/src/deepseek.ts:3

#Extends

OpenAICompatibleConfig

#Properties

#apiKey

apiKey: string

Defined in: adapters/src/openai.ts:6

#Inherited from

OpenAICompatibleConfig.apiKey

#baseUrl?

optional baseUrl?: string

Defined in: adapters/src/openai.ts:8

#Inherited from

OpenAICompatibleConfig.baseUrl

#capabilities?

optional capabilities?: Partial<AdapterCapabilities>

Defined in: adapters/src/openai.ts:19

Explicit capability facts for OpenAI-compatible providers with custom models.

#Inherited from

OpenAICompatibleConfig.capabilities

#includeUsage?

optional includeUsage?: boolean

Defined in: adapters/src/openai.ts:17

#Inherited from

OpenAICompatibleConfig.includeUsage

#model

model: string

Defined in: adapters/src/openai.ts:7

#Inherited from

OpenAICompatibleConfig.model

#retry?

optional retry?: RetryOptions

Defined in: adapters/src/openai.ts:9

#Inherited from

OpenAICompatibleConfig.retry

#Interface: DeepSeekEmbedderConfig

Defined in: adapters/src/embedders/deepseek.ts:3

#Extends

OpenAICompatibleEmbedderConfig

#Properties

#apiKey

apiKey: string

Defined in: adapters/src/embedders/openai-compatible.ts:6

#Inherited from

OpenAICompatibleEmbedderConfig.apiKey

#baseUrl?

optional baseUrl?: string

Defined in: adapters/src/embedders/openai-compatible.ts:8

#Inherited from

OpenAICompatibleEmbedderConfig.baseUrl

#model

model: string

Defined in: adapters/src/embedders/openai-compatible.ts:7

#Inherited from

OpenAICompatibleEmbedderConfig.model

#Interface: DeprecationPolicy

Defined in: adapters/src/deprecation.ts:32

#Properties

#logger?

optional logger?: (msg) => void

Defined in: adapters/src/deprecation.ts:36

Sink for warnings. Defaults to console.warn.

#Parameters

msg

string

#Returns

void

#onDeprecation

onDeprecation: DeprecationAction

Defined in: adapters/src/deprecation.ts:33

#table?

optional table?: ModelDeprecation[]

Defined in: adapters/src/deprecation.ts:34

#Interface: EnsembleBranchResult

Defined in: adapters/src/ensemble.ts:12

#Properties

#chunks

chunks: StreamChunk[]

Defined in: adapters/src/ensemble.ts:15

#error?

optional error?: Error

Defined in: adapters/src/ensemble.ts:16

#id

id: string

Defined in: adapters/src/ensemble.ts:13

#text

text: string

Defined in: adapters/src/ensemble.ts:14

#Interface: EnsembleCandidate

Defined in: adapters/src/ensemble.ts:5

#Properties

#adapter

adapter: AdapterFactory

Defined in: adapters/src/ensemble.ts:7

#id

id: string

Defined in: adapters/src/ensemble.ts:6

#weight?

optional weight?: number

Defined in: adapters/src/ensemble.ts:9

Weight used by 'weighted-vote' aggregator. Default 1.

#Interface: EnsembleOptions

Defined in: adapters/src/ensemble.ts:25

#Properties

#aggregate?

optional aggregate?: EnsembleAggregator

Defined in: adapters/src/ensemble.ts:28

How to combine branches into the single output text. Default 'majority-vote'.

#candidates

candidates: EnsembleCandidate[]

Defined in: adapters/src/ensemble.ts:26

#onBranches?

optional onBranches?: (branches) => void

Defined in: adapters/src/ensemble.ts:32

Observability hook — fires once with every branch's result.

#Parameters

branches

EnsembleBranchResult[]

#Returns

void

#timeoutMs?

optional timeoutMs?: number

Defined in: adapters/src/ensemble.ts:30

Per-candidate timeout in ms. Branches that time out are marked with an error.

#Interface: FallbackCandidate

Defined in: adapters/src/fallback.ts:15

#Properties

#adapter

adapter: AdapterFactory

Defined in: adapters/src/fallback.ts:17

#id

id: string

Defined in: adapters/src/fallback.ts:16

#Interface: FallbackOptions

Defined in: adapters/src/fallback.ts:5

#Properties

#onFallback?

optional onFallback?: (from) => void

Defined in: adapters/src/fallback.ts:12

Observability hook — fires when one adapter fails and the chain advances.

#Parameters

from

error

Error

id

string

index

number

#Returns

void

#shouldRetry?

optional shouldRetry?: (error, index) => boolean

Defined in: adapters/src/fallback.ts:10

Predicate deciding whether an error from a given adapter should trigger fall-through to the next. Default: always retry the next.

#Parameters

error

Error

index

number

#Returns

boolean

#Interface: FireworksConfig

Defined in: adapters/src/fireworks.ts:3

#Extends

OpenAICompatibleConfig

#Properties

#apiKey

apiKey: string

Defined in: adapters/src/openai.ts:6

#Inherited from

OpenAICompatibleConfig.apiKey

#baseUrl?

optional baseUrl?: string

Defined in: adapters/src/openai.ts:8

#Inherited from

OpenAICompatibleConfig.baseUrl

#capabilities?

optional capabilities?: Partial<AdapterCapabilities>

Defined in: adapters/src/openai.ts:19

Explicit capability facts for OpenAI-compatible providers with custom models.

#Inherited from

OpenAICompatibleConfig.capabilities

#includeUsage?

optional includeUsage?: boolean

Defined in: adapters/src/openai.ts:17

#Inherited from

OpenAICompatibleConfig.includeUsage

#model

model: string

Defined in: adapters/src/openai.ts:7

#Inherited from

OpenAICompatibleConfig.model

#retry?

optional retry?: RetryOptions

Defined in: adapters/src/openai.ts:9

#Inherited from

OpenAICompatibleConfig.retry

#Interface: GeminiConfig

Defined in: adapters/src/gemini.ts:6

#Properties

#apiKey

apiKey: string

Defined in: adapters/src/gemini.ts:7

#baseUrl?

optional baseUrl?: string

Defined in: adapters/src/gemini.ts:9

#model

model: string

Defined in: adapters/src/gemini.ts:8

#retry?

optional retry?: RetryOptions

Defined in: adapters/src/gemini.ts:10

#Interface: GeminiEmbedderConfig

Defined in: adapters/src/embedders/gemini.ts:4

#Properties

#apiKey

apiKey: string

Defined in: adapters/src/embedders/gemini.ts:5

#baseUrl?

optional baseUrl?: string

Defined in: adapters/src/embedders/gemini.ts:7

#model?

optional model?: string

Defined in: adapters/src/embedders/gemini.ts:6

#Interface: GenericAdapterConfig

Defined in: adapters/src/types.ts:20

#Properties

#send

send: (request, signal?) => Promise<ReadableStream<any>>

Defined in: adapters/src/types.ts:21

#Parameters

request

AdapterRequest

signal?

AbortSignal

#Returns

Promise<ReadableStream<any>>

#Interface: GrokConfig

Defined in: adapters/src/grok.ts:3

#Extends

OpenAICompatibleConfig

#Properties

#apiKey

apiKey: string

Defined in: adapters/src/openai.ts:6

#Inherited from

OpenAICompatibleConfig.apiKey

#baseUrl?

optional baseUrl?: string

Defined in: adapters/src/openai.ts:8

#Inherited from

OpenAICompatibleConfig.baseUrl

#capabilities?

optional capabilities?: Partial<AdapterCapabilities>

Defined in: adapters/src/openai.ts:19

Explicit capability facts for OpenAI-compatible providers with custom models.

#Inherited from

OpenAICompatibleConfig.capabilities

#includeUsage?

optional includeUsage?: boolean

Defined in: adapters/src/openai.ts:17

#Inherited from

OpenAICompatibleConfig.includeUsage

#model

model: string

Defined in: adapters/src/openai.ts:7

#Inherited from

OpenAICompatibleConfig.model

#retry?

optional retry?: RetryOptions

Defined in: adapters/src/openai.ts:9

#Inherited from

OpenAICompatibleConfig.retry

#Interface: GrokEmbedderConfig

Defined in: adapters/src/embedders/grok.ts:3

#Extends

OpenAICompatibleEmbedderConfig

#Properties

#apiKey

apiKey: string

Defined in: adapters/src/embedders/openai-compatible.ts:6

#Inherited from

OpenAICompatibleEmbedderConfig.apiKey

#baseUrl?

optional baseUrl?: string

Defined in: adapters/src/embedders/openai-compatible.ts:8

#Inherited from

OpenAICompatibleEmbedderConfig.baseUrl

#model

model: string

Defined in: adapters/src/embedders/openai-compatible.ts:7

#Inherited from

OpenAICompatibleEmbedderConfig.model

#Interface: GroqConfig

Defined in: adapters/src/groq.ts:4

#Extends

OpenAICompatibleConfig

#Properties

#apiKey

apiKey: string

Defined in: adapters/src/openai.ts:6

#Inherited from

OpenAICompatibleConfig.apiKey

#baseUrl?

optional baseUrl?: string

Defined in: adapters/src/openai.ts:8

#Inherited from

OpenAICompatibleConfig.baseUrl

#capabilities?

optional capabilities?: Partial<AdapterCapabilities>

Defined in: adapters/src/openai.ts:19

Explicit capability facts for OpenAI-compatible providers with custom models.

#Inherited from

OpenAICompatibleConfig.capabilities

#includeUsage?

optional includeUsage?: boolean

Defined in: adapters/src/openai.ts:17

#Inherited from

OpenAICompatibleConfig.includeUsage

#model

model: string

Defined in: adapters/src/openai.ts:7

#Inherited from

OpenAICompatibleConfig.model

#retry?

optional retry?: RetryOptions

Defined in: adapters/src/openai.ts:9

#Inherited from

OpenAICompatibleConfig.retry

#Interface: HuggingFaceConfig

Defined in: adapters/src/huggingface.ts:3

#Extends

OpenAICompatibleConfig

#Properties

#apiKey

apiKey: string

Defined in: adapters/src/openai.ts:6

#Inherited from

OpenAICompatibleConfig.apiKey

#baseUrl?

optional baseUrl?: string

Defined in: adapters/src/openai.ts:8

#Inherited from

OpenAICompatibleConfig.baseUrl

#capabilities?

optional capabilities?: Partial<AdapterCapabilities>

Defined in: adapters/src/openai.ts:19

Explicit capability facts for OpenAI-compatible providers with custom models.

#Inherited from

OpenAICompatibleConfig.capabilities

#includeUsage?

optional includeUsage?: boolean

Defined in: adapters/src/openai.ts:17

#Inherited from

OpenAICompatibleConfig.includeUsage

#model

model: string

Defined in: adapters/src/openai.ts:7

#Inherited from

OpenAICompatibleConfig.model

#retry?

optional retry?: RetryOptions

Defined in: adapters/src/openai.ts:9

#Inherited from

OpenAICompatibleConfig.retry

#Interface: KimiConfig

Defined in: adapters/src/kimi.ts:3

#Extends

OpenAICompatibleConfig

#Properties

#apiKey

apiKey: string

Defined in: adapters/src/openai.ts:6

#Inherited from

OpenAICompatibleConfig.apiKey

#baseUrl?

optional baseUrl?: string

Defined in: adapters/src/openai.ts:8

#Inherited from

OpenAICompatibleConfig.baseUrl

#capabilities?

optional capabilities?: Partial<AdapterCapabilities>

Defined in: adapters/src/openai.ts:19

Explicit capability facts for OpenAI-compatible providers with custom models.

#Inherited from

OpenAICompatibleConfig.capabilities

#includeUsage?

optional includeUsage?: boolean

Defined in: adapters/src/openai.ts:17

#Inherited from

OpenAICompatibleConfig.includeUsage

#model

model: string

Defined in: adapters/src/openai.ts:7

#Inherited from

OpenAICompatibleConfig.model

#retry?

optional retry?: RetryOptions

Defined in: adapters/src/openai.ts:9

#Inherited from

OpenAICompatibleConfig.retry

#Interface: KimiEmbedderConfig

Defined in: adapters/src/embedders/kimi.ts:3

#Extends

OpenAICompatibleEmbedderConfig

#Properties

#apiKey

apiKey: string

Defined in: adapters/src/embedders/openai-compatible.ts:6

#Inherited from

OpenAICompatibleEmbedderConfig.apiKey

#baseUrl?

optional baseUrl?: string

Defined in: adapters/src/embedders/openai-compatible.ts:8

#Inherited from

OpenAICompatibleEmbedderConfig.baseUrl

#model

model: string

Defined in: adapters/src/embedders/openai-compatible.ts:7

#Inherited from

OpenAICompatibleEmbedderConfig.model

#Interface: LangChainConfig

Defined in: adapters/src/langchain.ts:9

#Properties

#mode?

optional mode?: "stream" | "events"

Defined in: adapters/src/langchain.ts:11

#runnable

runnable: LangChainRunnable

Defined in: adapters/src/langchain.ts:10

#Interface: LangGraphConfig

Defined in: adapters/src/langchain.ts:99

#Properties

#graph

graph: LangChainRunnable

Defined in: adapters/src/langchain.ts:100

#Interface: LlamaCppConfig

Defined in: adapters/src/llamacpp.ts:3

#Extends

OpenAICompatibleConfig

#Properties

#apiKey

apiKey: string

Defined in: adapters/src/openai.ts:6

#Inherited from

OpenAICompatibleConfig.apiKey

#baseUrl?

optional baseUrl?: string

Defined in: adapters/src/openai.ts:8

#Inherited from

OpenAICompatibleConfig.baseUrl

#capabilities?

optional capabilities?: Partial<AdapterCapabilities>

Defined in: adapters/src/openai.ts:19

Explicit capability facts for OpenAI-compatible providers with custom models.

#Inherited from

OpenAICompatibleConfig.capabilities

#includeUsage?

optional includeUsage?: boolean

Defined in: adapters/src/openai.ts:17

#Inherited from

OpenAICompatibleConfig.includeUsage

#model

model: string

Defined in: adapters/src/openai.ts:7

#Inherited from

OpenAICompatibleConfig.model

#retry?

optional retry?: RetryOptions

Defined in: adapters/src/openai.ts:9

#Inherited from

OpenAICompatibleConfig.retry

#Interface: LMStudioConfig

Defined in: adapters/src/lmstudio.ts:3

#Extends

OpenAICompatibleConfig

#Properties

#apiKey

apiKey: string

Defined in: adapters/src/openai.ts:6

#Inherited from

OpenAICompatibleConfig.apiKey

#baseUrl?

optional baseUrl?: string

Defined in: adapters/src/openai.ts:8

#Inherited from

OpenAICompatibleConfig.baseUrl

#capabilities?

optional capabilities?: Partial<AdapterCapabilities>

Defined in: adapters/src/openai.ts:19

Explicit capability facts for OpenAI-compatible providers with custom models.

#Inherited from

OpenAICompatibleConfig.capabilities

#includeUsage?

optional includeUsage?: boolean

Defined in: adapters/src/openai.ts:17

#Inherited from

OpenAICompatibleConfig.includeUsage

#model

model: string

Defined in: adapters/src/openai.ts:7

#Inherited from

OpenAICompatibleConfig.model

#retry?

optional retry?: RetryOptions

Defined in: adapters/src/openai.ts:9

#Inherited from

OpenAICompatibleConfig.retry

#Interface: MistralConfig

Defined in: adapters/src/mistral.ts:3

#Extends

OpenAICompatibleConfig

#Properties

#apiKey

apiKey: string

Defined in: adapters/src/openai.ts:6

#Inherited from

OpenAICompatibleConfig.apiKey

#baseUrl?

optional baseUrl?: string

Defined in: adapters/src/openai.ts:8

#Inherited from

OpenAICompatibleConfig.baseUrl

#capabilities?

optional capabilities?: Partial<AdapterCapabilities>

Defined in: adapters/src/openai.ts:19

Explicit capability facts for OpenAI-compatible providers with custom models.

#Inherited from

OpenAICompatibleConfig.capabilities

#includeUsage?

optional includeUsage?: boolean

Defined in: adapters/src/openai.ts:17

#Inherited from

OpenAICompatibleConfig.includeUsage

#model

model: string

Defined in: adapters/src/openai.ts:7

#Inherited from

OpenAICompatibleConfig.model

#retry?

optional retry?: RetryOptions

Defined in: adapters/src/openai.ts:9

#Inherited from

OpenAICompatibleConfig.retry

#Interface: MockAdapterOptions

Defined in: adapters/src/mock.ts:12

#Properties

#delayMs?

optional delayMs?: number

Defined in: adapters/src/mock.ts:19

ms between yielded chunks. Default 0 (synchronous).

#history?

optional history?: AdapterRequest[]

Defined in: adapters/src/mock.ts:21

Track every request the adapter received. Useful for assertions.

#response

response: MockResponse | MockResponse[]

Defined in: adapters/src/mock.ts:17

Static chunks, a request-aware function, or a sequence of responses (the i-th call returns the i-th item, looping when exhausted).

#Interface: ModelDeprecation

Defined in: adapters/src/deprecation.ts:21

#Properties

#model

model: string

Defined in: adapters/src/deprecation.ts:23

#note?

optional note?: string

Defined in: adapters/src/deprecation.ts:29

Optional human-readable reason / link.

#provider

provider: string

Defined in: adapters/src/deprecation.ts:22

#successor?

optional successor?: string

Defined in: adapters/src/deprecation.ts:27

Suggested successor — used when onDeprecation: 'remap'.

#sunsetOn?

optional sunsetOn?: string

Defined in: adapters/src/deprecation.ts:25

ISO date the provider's sunset takes effect.

#Interface: OllamaConfig

Defined in: adapters/src/ollama.ts:5

#Properties

#baseUrl?

optional baseUrl?: string

Defined in: adapters/src/ollama.ts:7

#model

model: string

Defined in: adapters/src/ollama.ts:6

#retry?

optional retry?: RetryOptions

Defined in: adapters/src/ollama.ts:8

#Interface: OllamaEmbedderConfig

Defined in: adapters/src/embedders/ollama.ts:4

#Properties

#baseUrl?

optional baseUrl?: string

Defined in: adapters/src/embedders/ollama.ts:6

#model?

optional model?: string

Defined in: adapters/src/embedders/ollama.ts:5

#Interface: OpenAICompatibleEmbedderConfig

Defined in: adapters/src/embedders/openai-compatible.ts:5

#Extended by

#Properties

#apiKey

apiKey: string

Defined in: adapters/src/embedders/openai-compatible.ts:6

#baseUrl?

optional baseUrl?: string

Defined in: adapters/src/embedders/openai-compatible.ts:8

#model

model: string

Defined in: adapters/src/embedders/openai-compatible.ts:7

#Interface: OpenAIConfig

Defined in: adapters/src/openai.ts:5

#Properties

#apiKey

apiKey: string

Defined in: adapters/src/openai.ts:6

#baseUrl?

optional baseUrl?: string

Defined in: adapters/src/openai.ts:8

#capabilities?

optional capabilities?: Partial<AdapterCapabilities>

Defined in: adapters/src/openai.ts:19

Explicit capability facts for OpenAI-compatible providers with custom models.

#includeUsage?

optional includeUsage?: boolean

Defined in: adapters/src/openai.ts:17

#model

model: string

Defined in: adapters/src/openai.ts:7

#retry?

optional retry?: RetryOptions

Defined in: adapters/src/openai.ts:9

#Interface: OpenAIEmbedderConfig

Defined in: adapters/src/embedders/openai.ts:4

#Properties

#apiKey

apiKey: string

Defined in: adapters/src/embedders/openai.ts:5

#baseUrl?

optional baseUrl?: string

Defined in: adapters/src/embedders/openai.ts:7

#model?

optional model?: string

Defined in: adapters/src/embedders/openai.ts:6

#Interface: OpenRouterConfig

Defined in: adapters/src/openrouter.ts:3

#Extends

OpenAICompatibleConfig

#Properties

#apiKey

apiKey: string

Defined in: adapters/src/openai.ts:6

#Inherited from

OpenAICompatibleConfig.apiKey

#baseUrl?

optional baseUrl?: string

Defined in: adapters/src/openai.ts:8

#Inherited from

OpenAICompatibleConfig.baseUrl

#capabilities?

optional capabilities?: Partial<AdapterCapabilities>

Defined in: adapters/src/openai.ts:19

Explicit capability facts for OpenAI-compatible providers with custom models.

#Inherited from

OpenAICompatibleConfig.capabilities

#includeUsage?

optional includeUsage?: boolean

Defined in: adapters/src/openai.ts:17

#Inherited from

OpenAICompatibleConfig.includeUsage

#model

model: string

Defined in: adapters/src/openai.ts:7

#Inherited from

OpenAICompatibleConfig.model

#retry?

optional retry?: RetryOptions

Defined in: adapters/src/openai.ts:9

#Inherited from

OpenAICompatibleConfig.retry

#Interface: RecordedTurn

Defined in: adapters/src/mock.ts:152

#Properties

#chunks

chunks: StreamChunk[]

Defined in: adapters/src/mock.ts:158

Every chunk yielded by the wrapped adapter.

#recordedAt

recordedAt: string

Defined in: adapters/src/mock.ts:154

ISO timestamp when this turn was recorded.

#request

request: AdapterRequest

Defined in: adapters/src/mock.ts:156

The request that produced this turn.

#Interface: RecordingSink

Defined in: adapters/src/mock.ts:163

#Methods

#push()

push(turn): void | Promise<void>

Defined in: adapters/src/mock.ts:164

#Parameters

turn

RecordedTurn

#Returns

void | Promise<void>

#Interface: ReplicateConfig

Defined in: adapters/src/replicate.ts:4

#Properties

#apiKey

apiKey: string

Defined in: adapters/src/replicate.ts:5

#baseUrl?

optional baseUrl?: string

Defined in: adapters/src/replicate.ts:11

Override the prediction endpoint base.

#model

model: string

Defined in: adapters/src/replicate.ts:7

Replicate model id, e.g. meta/meta-llama-3-70b-instruct.

#toInput?

optional toInput?: (request) => Record<string, unknown>

Defined in: adapters/src/replicate.ts:13

Map AdapterRequest → Replicate input object. Defaults to \{ prompt \}.

#Parameters

request

AdapterRequest

#Returns

Record<string, unknown>

#version?

optional version?: string

Defined in: adapters/src/replicate.ts:9

Optional pinned version hash (for non-official models).

#Interface: ResolveModelInput

Defined in: adapters/src/deprecation.ts:61

#Properties

#model

model: string

Defined in: adapters/src/deprecation.ts:63

#provider

provider: string

Defined in: adapters/src/deprecation.ts:62

#Interface: ResolveModelResult

Defined in: adapters/src/deprecation.ts:66

#Properties

#deprecation?

optional deprecation?: ModelDeprecation

Defined in: adapters/src/deprecation.ts:70

#model

model: string

Defined in: adapters/src/deprecation.ts:68

Final model id to use (may differ from input if remapped).

#remapped

remapped: boolean

Defined in: adapters/src/deprecation.ts:69

#Interface: RetryOptions

Defined in: adapters/src/utils.ts:360

Retry knobs for adapter fetches. Tunable per call to createStreamSource.

Default behavior:

3 attempts total (1 initial + 2 retries)
exponential backoff: 500ms, 1000ms, 2000ms ... (capped at maxDelayMs)
full jitter on each delay
retry on HTTP 408, 429, 500, 502, 503, 504
retry on network errors (fetch throws)
DO NOT retry on 4xx other than 408/429 (those are bad requests / auth)
retries only the initial fetch — never mid-stream
respects Retry-After header when present

#Properties

#baseDelayMs?

optional baseDelayMs?: number

Defined in: adapters/src/utils.ts:362

#jitter?

optional jitter?: boolean

Defined in: adapters/src/utils.ts:364

#maxAttempts?

optional maxAttempts?: number

Defined in: adapters/src/utils.ts:361

#maxDelayMs?

optional maxDelayMs?: number

Defined in: adapters/src/utils.ts:363

#onRetry?

optional onRetry?: (info) => void

Defined in: adapters/src/utils.ts:367

Hook for tests + logging. Called after every failed attempt.

#Parameters

info

attempt

number

delayMs

number

reason

string

#Returns

void

#retryOn?

optional retryOn?: (info) => boolean

Defined in: adapters/src/utils.ts:365

#Parameters

info

attempt

number

error?

unknown

response?

Response

#Returns

boolean

#sleep?

optional sleep?: (ms) => Promise<void>

Defined in: adapters/src/utils.ts:369

Sleep override for tests. Defaults to setTimeout.

#Parameters

ms

number

#Returns

Promise<void>

#Interface: RotatingCredentials

Defined in: adapters/src/credential-rotation.ts:21

#Properties

#current

current: CredentialResolver

Defined in: adapters/src/credential-rotation.ts:23

Resolve the current secret. Opt-in adapters call this on every request.

#onRotate

onRotate: (handler) => () => void

Defined in: adapters/src/credential-rotation.ts:27

Subscribe to rotation events. Returns an unsubscribe handle.

#Parameters

handler

(event) => void

#Returns

() => void

#rotate

rotate: (next) => Promise<string>

Defined in: adapters/src/credential-rotation.ts:25

Replace the in-memory secret. Returns the new value.

#Parameters

string

#Returns

Promise<string>

#Interface: RouterCandidate

Defined in: adapters/src/router.ts:12

#Properties

#adapter

adapter: AdapterFactory

Defined in: adapters/src/router.ts:14

#capabilities?

optional capabilities?: AdapterCapabilities

Defined in: adapters/src/router.ts:29

Capability override. Defaults to adapter.capabilities. Used to reject candidates missing required features (e.g. tools, multiModal).

#cost?

optional cost?: number

Defined in: adapters/src/router.ts:21

Relative cost hint (lower wins for policy='cheapest'). Only relative values matter — use $/1M tokens or any consistent unit.

#gCO2PerKtok?

optional gCO2PerKtok?: number

Defined in: adapters/src/router.ts:38

Estimated grid CO2 intensity (gCO2eq per 1k tokens) for this adapter+region. Lower wins for policy='greenest'. Use applyCarbonTable() to populate from DEFAULT_CARBON_TABLE or a custom table.

#id

id: string

Defined in: adapters/src/router.ts:13

#latencyMs?

optional latencyMs?: number

Defined in: adapters/src/router.ts:23

Typical latency in ms. Lower wins for policy='fastest'.

#region?

optional region?: DataRegion

Defined in: adapters/src/router.ts:16

Data residency region for this adapter endpoint.

#tags?

optional tags?: string[]

Defined in: adapters/src/router.ts:31

Free-form tags used by classifier routing (e.g. 'fast', 'coding').

#Interface: RouterOptions

Defined in: adapters/src/router.ts:49

#Properties

#candidates

candidates: RouterCandidate[]

Defined in: adapters/src/router.ts:50

#classify?

optional classify?: (request) => string | string[] | undefined

Defined in: adapters/src/router.ts:61

Fast path: inspect the request, return a candidate id or tag(s). Return undefined to fall back to policy.

#Parameters

request

AdapterRequest

#Returns

string | string[] | undefined

#onRoute?

optional onRoute?: (decision) => void

Defined in: adapters/src/router.ts:63

Observability hook — fires once per decision.

#Parameters

decision

id

string

reason

string

request

AdapterRequest

#Returns

void

#policy?

optional policy?: RouterPolicy

Defined in: adapters/src/router.ts:56

Policy when classify doesn't pick a candidate. Default 'cheapest'.

#region?

optional region?: DataRegion

Defined in: adapters/src/router.ts:52

Require all selected adapters to match this data-residency region.

#regionOf?

optional regionOf?: (request) => DataRegion | undefined

Defined in: adapters/src/router.ts:54

Dynamic region selector. Takes precedence over region.

#Parameters

request

AdapterRequest

#Returns

DataRegion | undefined

#Interface: TogetherConfig

Defined in: adapters/src/together.ts:3

#Extends

OpenAICompatibleConfig

#Properties

#apiKey

apiKey: string

Defined in: adapters/src/openai.ts:6

#Inherited from

OpenAICompatibleConfig.apiKey

#baseUrl?

optional baseUrl?: string

Defined in: adapters/src/openai.ts:8

#Inherited from

OpenAICompatibleConfig.baseUrl

#capabilities?

optional capabilities?: Partial<AdapterCapabilities>

Defined in: adapters/src/openai.ts:19

Explicit capability facts for OpenAI-compatible providers with custom models.

#Inherited from

OpenAICompatibleConfig.capabilities

#includeUsage?

optional includeUsage?: boolean

Defined in: adapters/src/openai.ts:17

#Inherited from

OpenAICompatibleConfig.includeUsage

#model

model: string

Defined in: adapters/src/openai.ts:7

#Inherited from

OpenAICompatibleConfig.model

#retry?

optional retry?: RetryOptions

Defined in: adapters/src/openai.ts:9

#Inherited from

OpenAICompatibleConfig.retry

#Interface: VercelAIConfig

Defined in: adapters/src/vercel-ai.ts:7

#Properties

#api

api: string

Defined in: adapters/src/vercel-ai.ts:8

#headers?

optional headers?: Record<string, string>

Defined in: adapters/src/vercel-ai.ts:9

#retry?

optional retry?: RetryOptions

Defined in: adapters/src/vercel-ai.ts:10

#Interface: VertexConfig

Defined in: adapters/src/vertex.ts:6

#Properties

#accessToken

accessToken: string | (() => string | Promise<string>)

Defined in: adapters/src/vertex.ts:19

OAuth2 access token, or a function returning one (called per request).

Use the google-auth-library (or gcloud auth print-access-token) to mint these — the adapter intentionally doesn't take that as a hard dep.

#model

model: string

Defined in: adapters/src/vertex.ts:12

Vertex model id, e.g. gemini-2.5-pro.

#project

project: string

Defined in: adapters/src/vertex.ts:8

GCP project id.

#publisher?

optional publisher?: string

Defined in: adapters/src/vertex.ts:21

Override the publisher (defaults to google; set for partner publishers).

#region

region: string

Defined in: adapters/src/vertex.ts:10

Region, e.g. us-central1.

#retry?

optional retry?: RetryOptions

Defined in: adapters/src/vertex.ts:22

#Interface: VLLMConfig

Defined in: adapters/src/vllm.ts:3

#Extends

OpenAICompatibleConfig

#Properties

#apiKey

apiKey: string

Defined in: adapters/src/openai.ts:6

#Inherited from

OpenAICompatibleConfig.apiKey

#baseUrl?

optional baseUrl?: string

Defined in: adapters/src/openai.ts:8

#Inherited from

OpenAICompatibleConfig.baseUrl

#capabilities?

optional capabilities?: Partial<AdapterCapabilities>

Defined in: adapters/src/openai.ts:19

Explicit capability facts for OpenAI-compatible providers with custom models.

#Inherited from

OpenAICompatibleConfig.capabilities

#includeUsage?

optional includeUsage?: boolean

Defined in: adapters/src/openai.ts:17

#Inherited from

OpenAICompatibleConfig.includeUsage

#model

model: string

Defined in: adapters/src/openai.ts:7

#Inherited from

OpenAICompatibleConfig.model

#retry?

optional retry?: RetryOptions

Defined in: adapters/src/openai.ts:9

#Inherited from

OpenAICompatibleConfig.retry

#Interface: WebLlmConfig

Defined in: adapters/src/webllm.ts:15

Browser-only adapter backed by WebLLM (https://github.com/mlc-ai/web-llm). Models run on-device via WebGPU; no network for inference. The MLCEngine is loaded lazily on first stream so apps can ship the import without paying the wasm cost up front.

@mlc-ai/web-llm is an optional peer dependency — install it alongside this package when you opt into browser-only inference.

#Properties

#engine?

optional engine?: WebLlmEngineLike

Defined in: adapters/src/webllm.ts:22

Override the engine to inject a pre-loaded one (the MLCEngine spin-up is non-trivial — apps usually warm it once, not per turn).

#model

model: string

Defined in: adapters/src/webllm.ts:17

Model id from MLC's catalog, e.g. Llama-3.1-8B-Instruct-q4f16_1-MLC.

#onProgress?

optional onProgress?: (info) => void

Defined in: adapters/src/webllm.ts:24

Engine progress callback (model download / compile percent).

#Returns

void

#Interface: WebLlmEngineLike

Defined in: adapters/src/webllm.ts:27

#Properties

#chat

chat: object

Defined in: adapters/src/webllm.ts:29

#completions

completions: object

completions.create()

create(params): AsyncIterable<{ choices: object[]; }, any, any> | Promise<AsyncIterable<{ choices: object[]; }, any, any>>

#Methods

#reload()

reload(model, opts?): Promise<void>

Defined in: adapters/src/webllm.ts:28

#Returns

Promise<void>

Explore nearby

✎ Edit this page on GitHub·Found a problem? Open an issue →·How to contribute →