@agentskit/adapters β Functions
API functions for @agentskit/adapters.
#Function: anthropic()
anthropic(
config):AdapterFactory
Defined in: packages/adapters/src/anthropic.ts:25
Creates an adapter for Anthropic Messages API.
#Parameters
#config
Adapter configuration.
#Returns
AdapterFactory
An AgentsKit adapter factory.
#Example
const adapter = anthropic({ apiKey: 'β¦', model: 'model-name' })#Function: applyCarbonTable()
applyCarbonTable(
candidates,options?):RouterCandidate[]
Defined in: packages/adapters/src/carbon.ts:105
Decorate a list of router candidates with gCO2PerKtok from a
carbon table. Returns a new list β does not mutate the input.
const candidates = applyCarbonTable([
{ id: 'openai-eu', adapter: openai({...}), region: 'swedencentral', cost: 0.6 },
{ id: 'anthropic-us', adapter: anthropic({...}), region: 'us-east-1', cost: 0.5 },
])
const router = createRouter({ candidates, policy: 'green-cost' })#Parameters
#candidates
#options?
ApplyCarbonOptions = \{\}
#Returns
#Function: azureOpenAI()
azureOpenAI(
config):AdapterFactory
Defined in: packages/adapters/src/azure-openai.ts:31
Creates an adapter for Azure OpenAI chat completions.
#Parameters
#config
Adapter configuration.
#Returns
AdapterFactory
An AgentsKit adapter factory.
#Example
const adapter = azureOpenAI({ apiKey: 'β¦', endpoint: 'https://example.openai.azure.com', deployment: 'chat' })#Function: bail()
bail(
config):AdapterFactory
Defined in: packages/adapters/src/bail.ts:22
Alibaba Bailian (Qwen) via the DashScope OpenAI-compatibility endpoint.
Supports the Qwen-2.5 / 3 chat series and Qwen-VL multimodal models. APAC users typically prefer this over OpenAI for latency + data residency.
Default model: qwen-max.
#Parameters
#config
Partial<BailConfig> & object
#Returns
AdapterFactory
#Function: bedrock()
bedrock(
config):AdapterFactory
Defined in: packages/adapters/src/bedrock.ts:201
Creates an adapter for Amazon Bedrock Converse streaming.
#Parameters
#config
Adapter configuration.
#Returns
AdapterFactory
An AgentsKit adapter factory.
#Example
const adapter = bedrock({ model: 'anthropic.claude-3-5-sonnet-20241022-v2:0' })#Function: cerebras()
cerebras(
config):AdapterFactory
Defined in: packages/adapters/src/cerebras.ts:20
Cerebras β ultra-fast inference on wafer-scale chips. OpenAI-compatible endpoint serving Llama / Qwen models with very low first-token latency.
Default model: llama-3.3-70b.
#Parameters
#config
Partial<CerebrasConfig> & object
#Returns
AdapterFactory
#Function: chunkText()
chunkText(
text,targetSize?):string[]
Defined in: packages/adapters/src/utils.ts:363
Chunk-splitter that turns one large string into N streamable text chunks. Useful when a provider returns the full response in one shot and you want to feed it to a UI that expects streaming.
Default splits by whitespace boundaries with a target chunk size of ~32 characters.
#Parameters
#text
string
#targetSize?
number = 32
#Returns
string[]
#Function: cloudflareAiGateway()
cloudflareAiGateway(
config):AdapterFactory
Defined in: packages/adapters/src/cloudflare.ts:12
OpenAI-compatible AI Gateway endpoint; supports gateway-only authentication.
#Parameters
#config
#Returns
AdapterFactory
#Function: cohere()
cohere(
config):AdapterFactory
Defined in: packages/adapters/src/cohere.ts:26
Cohere Command models via Cohere's OpenAI-compatibility endpoint.
- Streams tokens via SSE (OpenAI-compatible chunks).
- Supports tool calls in the OpenAI
toolsshape. - Reports
usageon the final stream chunk when the upstream model returns it (Cohere's compatibility layer mirrors OpenAI'sstream_options: \{ include_usage: true \}semantics). - Inherits auto-retry from the shared OpenAI core (
retry).
Default model: command-r-plus. Override via model.
#Parameters
#config
Partial<CohereConfig> & object
#Returns
AdapterFactory
#Function: createAdapter()
createAdapter(
config):AdapterFactory
Defined in: packages/adapters/src/createAdapter.ts:10
Creates an adapter from send and stream parsing functions.
#Parameters
#config
Adapter configuration.
#Returns
AdapterFactory
An AgentsKit adapter factory.
#Function: createEnsembleAdapter()
createEnsembleAdapter(
options):AdapterFactory
Defined in: packages/adapters/src/ensemble.ts:111
Build an AdapterFactory that runs the same request against N
candidates in parallel, then aggregates the results into a single
text output. Unlike speculate (which picks a winner), ensemble
combines β majority vote, concatenation, longest, or a custom fn.
The returned source emits a single \{ type: 'text' \} chunk with
the aggregated output followed by \{ type: 'done' \}, so it plugs
into any runtime that expects a regular streaming adapter.
#Parameters
#options
#Returns
AdapterFactory
#Function: createFallbackAdapter()
createFallbackAdapter(
candidates,options?):AdapterFactory
Defined in: packages/adapters/src/fallback.ts:36
Try adapters in order. If the first fails (throws while opening, or errors mid-stream before emitting any non-done chunk), fall through to the next. As soon as a candidate produces its first real chunk, that's the committed one β we don't retroactively retry mid-stream.
Errors that happen after committing are propagated; the caller sees a normal streaming failure, not a mysterious cross-candidate retry that could duplicate tool calls.
#Parameters
#candidates
#options?
FallbackOptions = \{\}
#Returns
AdapterFactory
#Function: createModelProvider()
createModelProvider(
config):AdapterFactory
Defined in: packages/adapters/src/provider.ts:30
Single port over OpenRouter, Anthropic, OpenAI and CLIProxyAPI.
#Parameters
#config
#Returns
AdapterFactory
#Function: createOpenAICompatibleEmbedder()
createOpenAICompatibleEmbedder(
provider,defaultBaseUrl): (config) =>EmbedFn
Defined in: packages/adapters/src/embedders/openai-compatible.ts:55
Creates an embedder for an OpenAI-compatible embeddings API.
#Parameters
#provider
string
Value passed to the function.
#defaultBaseUrl
string
Value passed to the function.
#Returns
The the function result result.
(config) => EmbedFn
#Function: createRotatingCredentials()
createRotatingCredentials(
initial,options):RotatingCredentials
Defined in: packages/adapters/src/credential-rotation.ts:61
Creates a credential provider that refreshes expiring credentials.
#Parameters
#initial
string
Value passed to the function.
#options
Adapter configuration.
#id
string
#Returns
The RotatingCredentials result.
#Function: createRouter()
createRouter(
options):AdapterFactory
Defined in: packages/adapters/src/router.ts:136
Build an AdapterFactory that picks one of N candidates per request.
Resolution order:
classify(request)returns a candidate id β use itclassify(request)returns tag(s) β filter by tags, thenpolicy- Fall back to
policyover all capability-matched candidates
#Parameters
#options
#Returns
AdapterFactory
#Function: estimateCO2Grams()
estimateCO2Grams(
gCO2PerKtok,tokens):number
Defined in: packages/adapters/src/carbon.ts:125
Estimated CO2 emitted by a single completion (grams). Multiply
gCO2PerKtok by tokens / 1000. Provided as a convenience for
dashboards / chargeback reports β runtimes typically derive this
from token usage events.
#Parameters
#gCO2PerKtok
number | undefined
#tokens
number
#Returns
number
#Function: fetchWithRetry()
fetchWithRetry(
doFetch,signal,retryOpt?):Promise<Response>
Defined in: packages/adapters/src/retry.ts:67
Run doFetch with retries; return the final response on status exhaustion
and propagate the final transport error. Only the initial fetch retries;
errors from onRetry or custom sleep stop the operation.
#Parameters
#doFetch
(signal) => Promise<Response>
Called once per attempt with the caller's abort signal.
#signal
AbortSignal
Caller-owned; abort rejects with AbortError.
#retryOpt?
RetryOptions = \{\}
Defaults to 3 attempts, 500/8000 ms, full jitter, and statuses 408/429/500/502/503/504.
#Returns
Promise<Response>
The successful or final response.
#Deprecated
Use retry from @agentskit/net; keep this wrapper through at
least adapters 0.20.0 or 90 days after deprecation, whichever is later.
#Throws
The final transport error or AbortError.
#Throws
AK_CONFIG_INVALID for invalid maxAttempts.
#Example
`const response = await fetchWithRetry(signal => fetch(url, { signal }), controller.signal)`#Function: gemini()
gemini(
config):AdapterFactory
Defined in: packages/adapters/src/gemini.ts:24
Creates an adapter for Gemini streaming chat.
#Parameters
#config
Adapter configuration.
#Returns
AdapterFactory
An AgentsKit adapter factory.
#Example
const adapter = gemini({ apiKey: 'β¦', model: 'model-name' })#Function: geminiEmbedder()
geminiEmbedder(
config):EmbedFn
Defined in: packages/adapters/src/embedders/gemini.ts:54
Creates an embedder for Gemini embeddings.
#Parameters
#config
Adapter configuration.
#Returns
EmbedFn
The EmbedFn result.
#Function: generic()
generic(
config):AdapterFactory
Defined in: packages/adapters/src/generic.ts:10
Creates an adapter from a generic request sender.
#Parameters
#config
Adapter configuration.
#Returns
AdapterFactory
An AgentsKit adapter factory.
#Function: groq()
groq(
config):AdapterFactory
Defined in: packages/adapters/src/groq.ts:20
Groq β OpenAI-compatible endpoint serving Llama / Mixtral on LPUs. Known for very low first-token latency.
Default model: openai/gpt-oss-120b.
#Parameters
#config
Partial<GroqConfig> & object
#Returns
AdapterFactory
#Function: inMemorySink()
inMemorySink():
RecordingSink&object
Defined in: packages/adapters/src/mock.ts:238
In-memory recording sink β useful for tests and ephemeral capture.
#Returns
RecordingSink & object
#Function: langchain()
langchain(
config):AdapterFactory
Defined in: packages/adapters/src/langchain.ts:35
Wraps a LangChain chat model as an AgentsKit adapter.
#Parameters
#config
Adapter configuration.
#Returns
AdapterFactory
An AgentsKit adapter factory.
#Function: langgraph()
langgraph(
config):AdapterFactory
Defined in: packages/adapters/src/langchain.ts:119
Wraps a LangGraph-compatible model as an AgentsKit adapter.
#Parameters
#config
Adapter configuration.
#Returns
AdapterFactory
An AgentsKit adapter factory.
#Function: mockAdapter()
mockAdapter(
options):AdapterFactory
Defined in: packages/adapters/src/mock.ts:68
A deterministic adapter for tests, demos, and dry-run experiments.
Conforms to ADR 0001 β Adapter contract:
- createSource is pure (A1) β no work until stream() runs
- Always emits a terminal chunk (A3)
- abort() is safe (A6)
- Does not mutate input messages (A7)
Examples:
// Static const adapter = mockAdapter({ response: [ { type: 'text', content: 'Hello!' }, { type: 'done' }, ], })
// Request-aware const adapter = mockAdapter({ response: req => { const last = req.messages[req.messages.length - 1]?.content ?? '' return [ { type: 'text', content: 'Echo: ' + last }, { type: 'done' }, ] }, })
// Sequenced β different output each call const adapter = mockAdapter({ response: [ [{ type: 'text', content: 'first' }, { type: 'done' }], [{ type: 'text', content: 'second' }, { type: 'done' }], ], })
#Parameters
#options
#Returns
AdapterFactory
#Function: ollama()
ollama(
config):AdapterFactory
Defined in: packages/adapters/src/ollama.ts:24
Creates an adapter for Ollama chat.
#Parameters
#config
Adapter configuration.
#Returns
AdapterFactory
An AgentsKit adapter factory.
#Example
const adapter = ollama({ model: 'llama3.1' })#Function: ollamaEmbedder()
ollamaEmbedder(
config):EmbedFn
Defined in: packages/adapters/src/embedders/ollama.ts:48
Creates an embedder for the Ollama embeddings API.
#Parameters
#config
Adapter configuration.
#Returns
EmbedFn
The EmbedFn result.
#Function: openai()
openai(
config):AdapterFactory
Defined in: packages/adapters/src/openai.ts:36
Creates an adapter for OpenAI chat completions.
#Parameters
#config
Adapter configuration.
#Returns
AdapterFactory
An AgentsKit adapter factory.
#Example
const adapter = openai({ apiKey: 'β¦', model: 'model-name' })#Function: openaiCompatible()
openaiCompatible(
config):AdapterFactory
Defined in: packages/adapters/src/openai-compatible.ts:22
An OpenAI-compatible endpoint with explicit transport configuration.
#Parameters
#config
OpenAICompatibleTransportConfig
#Returns
AdapterFactory
#Function: openaiEmbedder()
openaiEmbedder(
config):EmbedFn
Defined in: packages/adapters/src/embedders/openai.ts:52
Creates an embedder for the OpenAI embeddings API.
#Parameters
#config
Adapter configuration.
#Returns
EmbedFn
The EmbedFn result.
#Function: recordingAdapter()
recordingAdapter(
inner,sink):AdapterFactory
Defined in: packages/adapters/src/mock.ts:186
Wrap a real adapter so every turn is captured to a sink. Use this in dev to build up a fixture, then replay with replayAdapter() in tests.
#Parameters
#inner
AdapterFactory
#sink
#Returns
AdapterFactory
#Function: refreshCredentials()
refreshCredentials(
adapter,next,options?):Promise<boolean>
Defined in: packages/adapters/src/credential-rotation.ts:96
Refresh credentials on any object that implements
CredentialRefreshable. A no-op (with a debug log) if the adapter
doesn't support rotation, so callers can run the rotation playbook
across a heterogeneous set of adapters without branching.
Most stock adapters do not implement refreshCredentials β this
remains an opt-in primitive unless an adapter documents support.
#Parameters
#adapter
unknown
#next
string
#options?
#id?
string
#logger?
(msg) => void
#Returns
Promise<boolean>
#Function: replayAdapter()
replayAdapter(
fixture):AdapterFactory
Defined in: packages/adapters/src/mock.ts:252
Replay an adapter from a recorded fixture. Each turn maps 1:1 to a recorded entry by index β call N replays fixture[N % fixture.length].
#Parameters
#fixture
#Returns
AdapterFactory
#Function: replicate()
replicate(
config):AdapterFactory
Defined in: packages/adapters/src/replicate.ts:66
Creates an adapter for Replicate streaming predictions.
#Parameters
#config
Adapter configuration.
#Returns
AdapterFactory
An AgentsKit adapter factory.
#Example
const adapter = replicate({ apiKey: 'β¦', model: 'model-name' })#Function: resolveModel()
resolveModel(
input,policy):ResolveModelResult
Defined in: packages/adapters/src/deprecation.ts:91
Resolves a model against the deprecation table and policy.
#Parameters
#input
Input to resolve.
#policy
Policy used during resolution.
#Returns
The resolved model and any deprecation information.
#Function: simulateStream()
simulateStream(
doFetch,extractText,errorLabel,options?):StreamSource
Defined in: packages/adapters/src/utils.ts:390
Build a StreamSource from a non-streaming fetch. The adapter is auto-completing: it fetches once, then yields the text as a sequence of chunks so UIs see the same streaming shape they'd see from a native streaming provider.
Use this when you're wiring a provider that only has a non-streaming endpoint but you want consumers (useChat, the runtime) to get identical ergonomics.
#Parameters
#doFetch
(signal) => Promise<Response>
#extractText
(response) => Promise<string>
#errorLabel
string
#options?
#chunkSize?
number
#delayMs?
number
#retry?
#Returns
StreamSource
#Function: vercelAI()
vercelAI(
config):AdapterFactory
Defined in: packages/adapters/src/vercel-ai.ts:119
Creates an adapter for the Vercel AI SDK.
#Parameters
#config
Adapter configuration.
#Returns
AdapterFactory
An AgentsKit adapter factory.
#Example
const adapter = vercelAI({ api: 'https://example.com/chat' })#Function: vertex()
vertex(
config):AdapterFactory
Defined in: packages/adapters/src/vertex.ts:36
Creates an adapter for Vertex AI Gemini streaming.
#Parameters
#config
Adapter configuration.
#Returns
AdapterFactory
An AgentsKit adapter factory.
#Example
const adapter = vertex({ project: 'my-project', region: 'us-central1', model: 'gemini-2.5-pro' })#Function: webllm()
webllm(
config):AdapterFactory
Defined in: packages/adapters/src/webllm.ts:73
Creates an adapter for WebLLM chat generation.
#Parameters
#config
Adapter configuration.
#Returns
AdapterFactory
An AgentsKit adapter factory.
#Example
const adapter = webllm({ model: 'Llama-3.1-8B-Instruct-q4f16_1-MLC' })#Function: withDeprecationPolicy()
withDeprecationPolicy<
F>(factory,options):F
Defined in: packages/adapters/src/deprecation.ts:145
Wrap an AdapterFactory so that any model field on its requests is
checked against the deprecation policy at startup. Convenience for
adapters that don't want to plumb the check into their own code:
const adapter = withDeprecationPolicy(openai({ apiKey, model: 'gpt-3.5-turbo-0301' }), {
provider: 'openai',
onDeprecation: 'remap',
})For adapters with internal model state (most), prefer calling
resolveModel() inside the factory and substituting before the
first request β this wrapper is a fallback for opaque adapters.
#Type Parameters
#F
F extends AdapterFactory
#Parameters
#factory
F
#options
object & DeprecationPolicy
#Returns
F
#Function: workersAi()
workersAi(
config):AdapterFactory
Defined in: packages/adapters/src/cloudflare.ts:29
Workers AI's OpenAI-compatible REST endpoint.
#Parameters
#config
#Returns
AdapterFactory