@ax-llm/ax Generated TypeScript API reference. typescript api api/reference build/apidocs/README.md api @ax-llm/ax

@ax-llm/ax

Enumerations

EnumerationDescription
AxAIAnthropicModel-
AxAIAnthropicVertexModel-
AxAICohereEmbedModelCohere: Models for use in embeddings
AxAICohereModelCohere: Models for text generation
AxAIDeepSeekModelDeepSeek: Models for text generation
AxAIGoogleGeminiEmbedModel-
AxAIGoogleGeminiEmbedTypes-
AxAIGoogleGeminiModel-
AxAIGoogleGeminiSafetyCategory-
AxAIGoogleGeminiSafetyThreshold-
AxAIGrokEmbedModels-
AxAIGrokModel-
AxAIMistralEmbedModels-
AxAIMistralModel-
AxAIOpenAIEmbedModel-
AxAIOpenAIModel-
AxAIOpenAIResponsesModel-
AxAIRekaModel-
AxJSRuntimePermissionPermissions that can be granted to the RLM JS interpreter sandbox. By default all dangerous globals are blocked; users opt in via this enum.

Classes

ClassDescription
AxAgentPipeline-based coordinator. Every run walks the same static sequence:
AxAgentClarificationError-
AxAgentContextMap-
AxAgentProtocolCompletionSignal-
AxAI-
AxAIAnthropic-
AxAIAzureOpenAI-
AxAICohereCohere provider using Cohere’s OpenAI compatibility endpoint.
AxAIDeepSeekDeepSeek AI client implementation extending OpenAI base functionality Provides access to DeepSeek’s language models through OpenAI-compatible API
AxAIGoogleGemini-
AxAIGrok-
AxAIMistral-
AxAIOpenAI-
AxAIOpenAIBase-
AxAIOpenAIResponsesBase class for OpenAI AI services using the /v1/responses API endpoint
AxAIOpenAIResponsesBaseBase class for OpenAI AI services using the /v1/responses API endpoint
AxAIRefusalError-
AxAIReka-
AxAIServiceAbortedError-
AxAIServiceAuthenticationError-
AxAIServiceError-
AxAIServiceNetworkError-
AxAIServiceResponseError-
AxAIServiceStatusError-
AxAIServiceStreamTerminatedError-
AxAIServiceTimeoutError-
AxAssertionError-
AxBalancerBalancer that rotates through services.
AxBaseAI-
AxBaseOptimizerAbstract base class for optimizers that provides common functionality and standardized handling of AxOptimizerArgs
AxBestOfN-
AxBootstrapFewShotAbstract base class for optimizers that provides common functionality and standardized handling of AxOptimizerArgs
AxContentProcessingErrorError thrown when content processing/transformation fails.
AxDefaultCostTracker-
AxDockerSession-
AxEmbeddingAdapter-
AxFlowInterface for flows that can be tuned, executed, and used in compositions. Provides methods for building and executing complex AI workflows.
AxFluentFieldType-
AxFunctionError-
AxFunctionProcessor-
AxGen-
AxGenerateError-
AxGEPASingle-module GEPA (reflective prompt evolution with Pareto sampling)
AxGEPAComponentSelector-
AxJSRuntimeBrowser-compatible JavaScript interpreter for RLM using Web Workers. Creates persistent sessions where variables survive across execute() calls.
AxMCPClient-
AxMCPHTTPSSETransport-
AxMCPStreamableHTTPTransport-
AxMCPStreambleHTTPTransport-
AxMediaNotSupportedErrorError thrown when an AI provider doesn’t support a required media type.
AxMemory-
AxMockAIService-
AxMultiServiceRouter-
AxOptimizedProgramImpl-
AxProgram-
AxPromptTemplate-
AxProviderRouterMulti-provider router that automatically selects optimal AI providers and handles content processing.
AxRateLimiterTokenUsage-
AxRefine-
AxRefineError-
AxSignature-
AxSignatureBuilder-
AxStopFunctionCallException-
AxStreamingAssertionError-
AxSynthAxSynth generates synthetic training data.
AxTestPrompt-
AxTokenLimitError-

Interfaces

InterfaceDescription
AxAgentConfig-
AxAgenticInterface for agents that can be used as child agents. Provides methods to get the agent’s function definition and features.
AxAIAnthropicArgs-
AxAIAnthropicContentBlockDeltaEvent-
AxAIAnthropicContentBlockStartEvent-
AxAIAnthropicContentBlockStopEvent-
AxAIAnthropicErrorEvent-
AxAIAnthropicMessageDeltaEvent-
AxAIAnthropicMessageStartEvent-
AxAIAnthropicMessageStopEvent-
AxAIAnthropicPingEvent-
AxAIFeatures-
AxAIGoogleGeminiArgs-
AxAIGoogleGeminiOptionsTools-
AxAIGrokOptionsTools-
AxAIGrokSearchSource-
AxAIMemory-
AxAIMetricsInstruments-
AxAIOpenAIArgs-
AxAIOpenAIBaseArgs-
AxAIOpenAIResponseDelta-
AxAIOpenAIResponsesArgsReady-to-use implementation of the OpenAI Responses API client This class uses OpenAI’s /v1/responses API endpoint which supports text, image, and audio inputs
AxAIOpenAIResponsesCodeInterpreterToolCall-
AxAIOpenAIResponsesComputerToolCall-
AxAIOpenAIResponsesContentPartAddedEvent-
AxAIOpenAIResponsesContentPartDoneEvent-
AxAIOpenAIResponsesDefineFunctionTool-
AxAIOpenAIResponsesErrorEvent-
AxAIOpenAIResponsesFileSearchCallCompletedEvent-
AxAIOpenAIResponsesFileSearchCallInProgressEvent-
AxAIOpenAIResponsesFileSearchCallSearchingEvent-
AxAIOpenAIResponsesFileSearchToolCall-
AxAIOpenAIResponsesFunctionCallArgumentsDeltaEvent-
AxAIOpenAIResponsesFunctionCallArgumentsDoneEvent-
AxAIOpenAIResponsesFunctionCallItem-
AxAIOpenAIResponsesImageGenerationCallCompletedEvent-
AxAIOpenAIResponsesImageGenerationCallGeneratingEvent-
AxAIOpenAIResponsesImageGenerationCallInProgressEvent-
AxAIOpenAIResponsesImageGenerationCallPartialImageEvent-
AxAIOpenAIResponsesImageGenerationToolCall-
AxAIOpenAIResponsesInputAudioContentPart-
AxAIOpenAIResponsesInputFunctionCallItem-
AxAIOpenAIResponsesInputFunctionCallOutputItem-
AxAIOpenAIResponsesInputImageUrlContentPart-
AxAIOpenAIResponsesInputMessageItem-
AxAIOpenAIResponsesInputTextContentPart-
AxAIOpenAIResponsesLocalShellToolCall-
AxAIOpenAIResponsesMCPCallArgumentsDeltaEvent-
AxAIOpenAIResponsesMCPCallArgumentsDoneEvent-
AxAIOpenAIResponsesMCPCallCompletedEvent-
AxAIOpenAIResponsesMCPCallFailedEvent-
AxAIOpenAIResponsesMCPCallInProgressEvent-
AxAIOpenAIResponsesMCPListToolsCompletedEvent-
AxAIOpenAIResponsesMCPListToolsFailedEvent-
AxAIOpenAIResponsesMCPListToolsInProgressEvent-
AxAIOpenAIResponsesMCPToolCall-
AxAIOpenAIResponsesOutputItemAddedEvent-
AxAIOpenAIResponsesOutputItemDoneEvent-
AxAIOpenAIResponsesOutputMessageItem-
AxAIOpenAIResponsesOutputRefusalContentPart-
AxAIOpenAIResponsesOutputTextAnnotationAddedEvent-
AxAIOpenAIResponsesOutputTextContentPart-
AxAIOpenAIResponsesOutputTextDeltaEvent-
AxAIOpenAIResponsesOutputTextDoneEvent-
AxAIOpenAIResponsesReasoningDeltaEvent-
AxAIOpenAIResponsesReasoningDoneEvent-
AxAIOpenAIResponsesReasoningItem-
AxAIOpenAIResponsesReasoningSummaryDeltaEvent-
AxAIOpenAIResponsesReasoningSummaryDoneEvent-
AxAIOpenAIResponsesReasoningSummaryPart-
AxAIOpenAIResponsesReasoningSummaryPartAddedEvent-
AxAIOpenAIResponsesReasoningSummaryPartDoneEvent-
AxAIOpenAIResponsesReasoningSummaryTextDeltaEvent-
AxAIOpenAIResponsesReasoningSummaryTextDoneEvent-
AxAIOpenAIResponsesRefusalDeltaEvent-
AxAIOpenAIResponsesRefusalDoneEvent-
AxAIOpenAIResponsesRequest-
AxAIOpenAIResponsesResponse-
AxAIOpenAIResponsesResponseCompletedEvent-
AxAIOpenAIResponsesResponseCreatedEvent-
AxAIOpenAIResponsesResponseFailedEvent-
AxAIOpenAIResponsesResponseIncompleteEvent-
AxAIOpenAIResponsesResponseInProgressEvent-
AxAIOpenAIResponsesResponseQueuedEvent-
AxAIOpenAIResponsesStreamEventBase-
AxAIOpenAIResponsesToolCallBase-
AxAIOpenAIResponsesWebSearchCallCompletedEvent-
AxAIOpenAIResponsesWebSearchCallInProgressEvent-
AxAIOpenAIResponsesWebSearchCallSearchingEvent-
AxAIOpenAIResponsesWebSearchToolCall-
AxAIService-
AxAIServiceImpl-
AxAIServiceMetrics-
AxAPI-
AxAPIConfig-
AxAPIResponseMetadata-
AxAssertion-
AxBaseAIArgs-
AxBootstrapOptimizerOptions-
AxCodeRuntimeA code runtime that can create persistent sessions. Implement this interface for your target runtime (Node.js, browser, WASM, etc.).
AxCodeSessionA persistent code execution session. Variables persist across execute() calls.
AxCompileOptions-
AxContentProcessingServicesServices for converting unsupported content types to text or optimized formats
AxContextPolicyConfigPublic context policy for the Actor loop. Users choose replay style via preset and overall prompt budget via budget.
AxCostTracker-
AxCostTrackerOptions-
AxDockerContainer-
AxField-
AxFieldOptionsPer-field companion options for Standard Schema (zod/valibot/arktype) inputs and outputs. These encode ax-specific hints that schema libraries don’t represent natively.
AxFieldProcessor-
AxFieldType-
AxFlowableInterface for flows that can be tuned, executed, and used in compositions. Provides methods for building and executing complex AI workflows.
AxFlowBranchEvaluationDataData types for different AxFlow logging events
AxFlowCompleteDataData types for different AxFlow logging events
AxFlowDynamicContext-
AxFlowErrorDataData types for different AxFlow logging events
AxFlowExecutionPlan-
AxFlowExecutionPlanGroup-
AxFlowExecutionPlanStep-
AxFlowLoggerDataData types for different AxFlow logging events
AxFlowOptions-
AxFlowParallelGroupCompleteDataData types for different AxFlow logging events
AxFlowParallelGroupStartDataData types for different AxFlow logging events
AxFlowStartDataData types for different AxFlow logging events
AxFlowStateDependencyAnalysisBest-effort static dependency extraction for flow execute mappings.
AxFlowStepCompleteDataData types for different AxFlow logging events
AxFlowStepStartDataData types for different AxFlow logging events
AxFlowTypedSubContext-
AxFluentFieldInfo-
AxForwardable-
AxGenMetricsInstruments-
AxGEPAAdapter-
AxGEPABootstrapOptions-
AxGEPAEvaluationBatch-
AxGEPAOptimizationReportStructured optimization report
AxJudgeOptions-
AxMCPAnnotations-
AxMCPAudioContent-
AxMCPBaseAnnotated-
AxMCPBlobResourceContents-
AxMCPClientCapabilities-
AxMCPClientOptions-
AxMCPCompletionArgument-
AxMCPCompletionRequest-
AxMCPCompletionResult-
AxMCPEmbeddedResource-
AxMCPFunctionOverride-
AxMCPIcon-
AxMCPImageContent-
AxMCPImplementationInfo-
AxMCPInitializeParams-
AxMCPInitializeResult-
AxMCPJSONRPCErrorResponse-
AxMCPJSONRPCNotification-
AxMCPJSONRPCRequest-
AxMCPJSONRPCSuccessResponse-
AxMCPListRootsResult-
AxMCPOAuthOptions-
AxMCPPaginatedRequest-
AxMCPPrompt-
AxMCPPromptArgument-
AxMCPPromptGetResult-
AxMCPPromptMessage-
AxMCPPromptsListResult-
AxMCPResource-
AxMCPResourceLink-
AxMCPResourceReadResult-
AxMCPResourcesListResult-
AxMCPResourceTemplate-
AxMCPResourceTemplatesListResult-
AxMCPRoot-
AxMCPServerCapabilities-
AxMCPSSRFProtectionOptions-
AxMCPStreamableHTTPTransportOptions-
AxMCPTextContent-
AxMCPTextResourceContents-
AxMCPTool-
AxMCPToolCallParams-
AxMCPToolCallResult-
AxMCPToolsListResult-
AxMCPTransport-
AxMetricsConfig-
AxMultiProviderConfigConfiguration for multi-provider routing with fallback capabilities
AxOptimizableComponent-
AxOptimizationCheckpoint-
AxOptimizationProgress-
AxOptimizationStats-
AxOptimizedProgram-
AxOptimizerInterface for optimizing AI programs through automated prompt tuning.
AxOptimizerMetricsConfig-
AxOptimizerMetricsInstruments-
AxOptimizerResult-
AxParetoResult-
AxProgrammable-
AxProgramOptions-
AxPromptTemplateOptions-
AxRateLimiterTokenUsageOptions-
AxRLMConfigRLM configuration for AxAgent.
AxRoutingResultResult of the routing process including provider selection and processing information
AxRuntimePrimitive-
AxSamplePickerOptions-
AxSignatureConfig-
AxStepContextMutable context object that flows through the generation loop. Accessible to functions and step hooks, enabling per-step control over model, options, functions, and loop flow.
AxStreamingAssertion-
AxStreamingEvent-
AxSynthesizerInit-
AxSynthesizerOptions-
AxSynthExampleA single synthesized example with input and expected output.
AxSynthOptionsConfiguration for synthetic data generation.
AxSynthResultResult from a synthesis run.
AxTunable-
AxUsable-

Type Aliases

Type AliasDescription
AxAgentActorTurnCallback-
AxAgentActorTurnCallbackArgs-
AxAgentClarification-
AxAgentClarificationChoice-
AxAgentClarificationKind-
AxAgentCompletionProtocol-
AxAgentContextEvent-
AxAgentContextMapConfig-
AxAgentContextMapOperation-
AxAgentContextMapOptions-
AxAgentContextMapSnapshot-
AxAgentContextMapUpdateResult-
AxAgentContextPressure-
AxAgentContextStage-
AxAgentDemosDemo traces for AxAgent’s split architecture. Actor demos use the runtime code field (javascriptCode for JavaScript, <language>Code for other runtimes such as pythonCode). Responder demos use the agent’s output type + optional input fields.
AxAgentDiscoveryPromptState-
AxAgentEvalDataset-
AxAgentEvalFunctionCall-
AxAgentEvalPrediction-
AxAgentEvalTask-
AxAgentExecutorResultPayload-
AxAgentForwardOptionsForward options for AxAgent.forward(...). Extends the dsp-layer AxProgramForwardOptionsWithModels with agent-specific knobs that only make sense at the agent boundary (currently skills for one-shot preloading). Forward-time skills merge on top of init-time skills (forward overrides by id).
AxAgentFunction-
AxAgentFunctionCall-
AxAgentFunctionCallRecorder-
AxAgentFunctionCollection-
AxAgentFunctionExample-
AxAgentFunctionGroup-
AxAgentFunctionModuleMeta-
AxAgentGuidanceLogEntry-
AxAgentGuidancePayload-
AxAgentGuidanceState-
AxAgentIdentity-
AxAgentInputUpdateCallback-
AxAgentJudgeEvalInput-
AxAgentJudgeEvalOutput-
AxAgentJudgeInput-
AxAgentJudgeOptions-
AxAgentJudgeOutput-
AxAgentMemoriesSearchFnMemories search callback. Receives the raw search strings and the snapshot of inputs.memories already loaded for the current run (deduped by id, sorted). Use the second argument to skip work for entries that are already in scope — for example, filter your vector search by id NOT IN alreadyLoaded so you don’t re-fetch and the actor doesn’t pay tokens for duplicates. Returning already-loaded entries is still safe (the runtime dedupes by id) but wastes work.
AxAgentMemoryEntry-
AxAgentMemoryResult-
AxAgentOnContextEvent-
AxAgentOnFunctionCall-
AxAgentOptimizationTargetDescriptor-
AxAgentOptimizeOptions-
AxAgentOptimizeResult-
AxAgentOptimizeTarget-
AxAgentOptions-
AxAgentRecursionOptionsForward options forwarded to the AxGen spawned by each llmQuery(...) call.
AxAgentRecursiveExpensiveNode-
AxAgentRecursiveFunctionCall-
AxAgentRecursiveNodeRole-
AxAgentRecursiveStats-
AxAgentRecursiveTargetId-
AxAgentRecursiveTraceNode-
AxAgentRecursiveTurn-
AxAgentRecursiveUsage-
AxAgentRuntimeCompletionState-
AxAgentRuntimeExecutionContext-
AxAgentRuntimeInputState-
AxAgentSkillResult-
AxAgentSkillsPromptState-
AxAgentSkillsSearchFn-
AxAgentState-
AxAgentStateActionLogEntry-
AxAgentStateCheckpointState-
AxAgentStateExecutorModelState-
AxAgentStateRuntimeEntry-
AxAgentStreamingForwardOptions-
AxAgentStructuredClarification-
AxAgentTestCompletionPayload-
AxAgentTestResult-
AxAgentUsage-
AxAgentUsedMemoriesCallback-
AxAgentUsedMemory-
AxAgentUsedSkill-
AxAgentUsedSkillsCallback-
AxAIAnthropicChatError-
AxAIAnthropicChatRequest-
AxAIAnthropicChatRequestCacheParam-
AxAIAnthropicChatResponse-
AxAIAnthropicChatResponseDelta-
AxAIAnthropicConfig-
AxAIAnthropicEffortLevel-
AxAIAnthropicEffortLevelMapping-
AxAIAnthropicFunctionTool-
AxAIAnthropicOutputConfig-
AxAIAnthropicRequestTool-
AxAIAnthropicStopDetails-
AxAIAnthropicTaskBudget-
AxAIAnthropicThinkingConfig-
AxAIAnthropicThinkingTokenBudgetLevels-
AxAIAnthropicThinkingWire-
AxAIAnthropicWebSearchTool-
AxAIArgs-
AxAIAzureOpenAIArgs-
AxAIAzureOpenAIConfig-
AxAICohereArgs-
AxAICohereChatRequest-
AxAICohereChatRequestToolResults-
AxAICohereChatResponse-
AxAICohereChatResponseDelta-
AxAICohereChatResponseToolCalls-
AxAICohereConfigCohere: Model options for text generation
AxAICohereEmbedRequest-
AxAICohereEmbedResponse-
AxAIDeepSeekArgsArguments type for initializing DeepSeek AI instances
AxAIEmbedModels-
AxAIGoogleGeminiBatchEmbedRequestAxAIGoogleGeminiEmbedRequest: Structure for making an embedding request to the Google Gemini API.
AxAIGoogleGeminiBatchEmbedResponseAxAIGoogleGeminiEmbedResponse: Structure for handling responses from the Google Gemini API embedding requests.
AxAIGoogleGeminiCacheCreateRequestRequest to create a context cache in Vertex AI / Gemini API.
AxAIGoogleGeminiCacheResponseResponse from creating/getting a context cache.
AxAIGoogleGeminiCacheUpdateRequestRequest to update a context cache (e.g., extend TTL).
AxAIGoogleGeminiChatRequest-
AxAIGoogleGeminiChatResponse-
AxAIGoogleGeminiChatResponseDelta-
AxAIGoogleGeminiConfigAxAIGoogleGeminiConfig: Configuration options for Google Gemini API
AxAIGoogleGeminiContent-
AxAIGoogleGeminiContentPart-
AxAIGoogleGeminiGenerationConfig-
AxAIGoogleGeminiRetrievalConfig-
AxAIGoogleGeminiSafetySettings-
AxAIGoogleGeminiThinkingConfig-
AxAIGoogleGeminiThinkingLevel-
AxAIGoogleGeminiThinkingLevelMappingMaps thinkingTokenBudget string levels to Gemini 3+ thinkingLevel values. ’none’ is handled separately (disables thinking or maps to ‘minimal’ for Gemini 3+).
AxAIGoogleGeminiThinkingTokenBudgetLevels-
AxAIGoogleGeminiTool-
AxAIGoogleGeminiToolConfig-
AxAIGoogleGeminiToolFunctionDeclaration-
AxAIGoogleGeminiToolGoogleMaps-
AxAIGoogleGeminiToolGoogleSearchRetrieval-
AxAIGoogleVertexBatchEmbedRequestAxAIGoogleVertexBatchEmbedRequest: Structure for making an embedding request to the Google Vertex API.
AxAIGoogleVertexBatchEmbedResponseAxAIGoogleVertexBatchEmbedResponse: Structure for handling responses from the Google Vertex API embedding requests.
AxAIGrokArgs-
AxAIGrokChatRequest-
AxAIInputModelList-
AxAIMistralArgs-
AxAIMistralChatRequest-
AxAIModelCatalogAudioSupport-
AxAIModelCatalogFilter-
AxAIModelCatalogModel-
AxAIModelCatalogModelCapabilities-
AxAIModelCatalogModelType-
AxAIModelCatalogOptions-
AxAIModelCatalogProvider-
AxAIModelCatalogProviderName-
AxAIModelList-
AxAIModelListBase-
AxAIModels-
AxAIOpenAIAnnotation-
AxAIOpenAIChatRequest-
AxAIOpenAIChatResponse-
AxAIOpenAIChatResponseDelta-
AxAIOpenAIConfig-
AxAIOpenAIEmbedRequest-
AxAIOpenAIEmbedResponse-
AxAIOpenAILogprob-
AxAIOpenAIResponsesConfig-
AxAIOpenAIResponsesInputContentPart-
AxAIOpenAIResponsesInputItem-
AxAIOpenAIResponsesOutputItem-
AxAIOpenAIResponsesStreamEvent-
AxAIOpenAIResponsesToolCall-
AxAIOpenAIResponsesToolChoice-
AxAIOpenAIResponsesToolDefinition-
AxAIOpenAIUrlCitation-
AxAIOpenAIUsage-
AxAIRekaArgs-
AxAIRekaChatRequest-
AxAIRekaChatResponse-
AxAIRekaChatResponseDelta-
AxAIRekaConfig-
AxAIRekaUsage-
AxAIServiceActionOptions-
AxAIServiceModelType-
AxAIServiceOptionsRuntime options for AI service requests.
AxAnyAgentic-
AxAttempt-
AxAudioFormat-
AxAudioInput-
AxBalancerOptionsOptions for the balancer.
AxBestOfNOptions-
AxChatAudioConfig-
AxChatAudioOutput-
AxChatLogEntryA single chat round-trip entry: the full prompt sent to the model and its response, with normalized roles and inline XML formatting.
AxChatLogMessageA normalized chat message with standard roles: system, user, assistant, tool. Assistant content uses inline XML: <think> for reasoning, <tool_call> for tool invocations.
AxChatRequest-
AxChatResponse-
AxChatResponseFunctionCall-
AxChatResponseResult-
AxCheckpointLoadFn-
AxCheckpointSaveFn-
AxCitation-
AxCodeSessionSnapshot-
AxCodeSessionSnapshotEntry-
AxContextCacheInfoInformation about a context cache entry (returned after creation or lookup).
AxContextCacheOperationContext cache operation to be executed by the base AI service. Providers define these operations; AxBaseAI executes them via apiCall().
AxContextCacheOptionsOptions for explicit context caching (e.g., Gemini/Vertex context caching). Allows caching large prompt prefixes for cost savings and lower latency.
AxContextCacheRegistryExternal registry for persisting context cache metadata. Useful for serverless/short-lived processes where in-memory storage is lost.
AxContextCacheRegistryEntryEntry stored in the context cache registry. Used for persisting cache metadata across process restarts.
AxContextFieldInput-
AxContextFieldPromptConfig-
AxContextPolicyBudget-
AxContextPolicyPresetOpinionated context replay presets for the Actor loop.
AxDateRange-
AxDateRangeValue-
AxDebugChatResponseUsage-
AxDiscoveryTurnSummary-
AxEmbedRequest-
AxEmbedResponse-
AxErrorCategory-
AxEvaluateArgs-
AxExample-
AxExamples-
AxExecutorModelPolicy-
AxExecutorModelPolicyEntry-
AxFieldProcessorProcess-
AxFieldTemplateFn-
AxFieldValue-
AxFlowForwardOptions-
AxFlowLogData-
AxFlowLoggerFunctionFunction type for AxFlow logging
AxFlowState-
AxFlowTypedParallelBranch-
AxFunction-
AxFunctionCallRecordRecord of a single function call executed during a step.
AxFunctionCallTrace-
AxFunctionHandler-
AxFunctionJSONSchema-
AxFunctionProvider-
AxFunctionResult-
AxFunctionResultFormatter-
AxGenDeltaOut-
AxGenerateErrorDetails-
AxGenerateResult-
AxGenIn-
AxGenInput-
AxGenOut-
AxGenOutput-
AxGenStreamingOut-
AxGEPABatchEvaluation-
AxGEPABatchRow-
AxGEPAComponentBanditState-
AxGEPAComponentTarget-
AxGEPAEvaluationState-
AxGEPAReflectiveTuple-
AxGEPATraceSummary-
AxGEPATraceSummaryCall-
AxIField-
AxInputFunctionType-
AxJSRuntimeNodePermissionAllowlistFine-grained Node Permission Model allowlist. Scopes --allow-fs-* and gates additional --allow-* flags that aren’t covered by the high-level permission enum.
AxJSRuntimeOutputMode-
AxJSRuntimeResourceLimitsNode worker_threads resource limits passthrough.
AxJudgeForwardOptions-
AxLlmQueryBudgetStateBudget state for llmQuery calls. Uses a shared global object for cross-tree tracking plus per-agent local counters to prevent any single child from starving siblings.
AxLlmQueryPromptMode-
AxLoggerData-
AxLoggerFunction-
AxMCPCompletionReference-
AxMCPContent-
AxMCPFetchOptions-
AxMCPFunctionDescription-
AxMCPJSONRPCMessage-
AxMCPJSONRPCResponse-
AxMCPJSONSchema-
AxMCPLoggingLevel-
AxMCPMeta-
AxMCPProtocolVersion-
AxMCPSSRFProtectionContext-
AxMCPTokenSet-
AxMemoryData-
AxMemoryMessageValue-
AxMetricFn-
AxMetricFnArgs-
AxMockAIServiceConfig-
AxModelConfigConfiguration options for AI model behavior.
AxModelInfo-
AxModelInfoWithProvider-
AxModelUsage-
AxMultiMetricFn-
AxNamedProgramInstance-
AxOptimizableValidatorGeneric component-optimization surface that any AxProgrammable can implement.
AxOptimizeOptions-
AxOptimizerArgs-
AxOptimizerLoggerData-
AxOptimizerLoggerFunction-
AxProgramDemos-
AxProgramExamples-
AxProgramForwardOptions-
AxProgramForwardOptionsWithModels-
AxProgramStreamingForwardOptions-
AxProgramStreamingForwardOptionsWithModels-
AxProgramTrace-
AxProgramUsage-
AxPromptMetrics-
AxProviderMetadata-
AxRateLimiterFunction-
AxRefineOptions-
AxRefineStrategy-
AxRenderedPrompt-
AxResolvedContextPolicy-
AxResolvedExecutorModelPolicy-
AxResolvedExecutorModelPolicyEntry-
AxResultPickerFunction-
AxResultPickerFunctionFieldResults-
AxResultPickerFunctionFunctionResults-
AxRewardFn-
AxRewardFnArgs-
AxRolloutTrace-
AxRuntimeCallableFormatArgs-
AxRuntimeLanguageInfo-
AxRuntimePrimitiveExample-
AxRuntimePrimitiveOverrideMap-
AxRuntimePrimitiveSignature-
AxRuntimePrimitiveStageRuntime primitive registry.
AxSelfTuningConfigConfiguration for LLM self-tuning capabilities. When enabled, an adjustGeneration function is auto-injected that lets the LLM adjust its own generation parameters.
AxSerializedOptimizedProgram-
AxSetExamplesOptions-
AxSignatureInput-
AxSpeechConfig-
AxSpeechRequest-
AxSpeechResponse-
AxStageDefinitionBuildOptions-
AxStageOptions-
AxStepHooksHooks called at various points during the multi-step generation loop.
AxStepUsageAccumulated token usage across steps.
AxStreamingFieldProcessorProcess-
AxSynthesizerRole-
AxThoughtBlockItemSingle thinking block item with its signature
AxTokenUsage-
AxTranscriptionRequest-
AxTranscriptionResponse-
AxTranscriptionSegment-
AxTypedExample-
AxWorkerRuntimeConfig-

Variables

VariableDescription
axAIAzureOpenAIBestConfig-
axAIAzureOpenAICreativeConfig-
axAIAzureOpenAIDefaultConfig-
axAIAzureOpenAIFastConfig-
axDefaultFlowLoggerDefault AxFlow logger with colors
axDefaultMetricsConfig-
axDefaultOptimizerLoggerDefault optimizer logger instance with color formatting
axDefaultOptimizerMetricsConfig-
AxEvalUtil-
axGlobals-
axModelInfoAnthropic-
axModelInfoCohere-
axModelInfoDeepSeek-
axModelInfoGoogleGeminiAxAIGoogleGemini: Model information
axModelInfoGrok-
axModelInfoMistral-
axModelInfoOpenAIOpenAI: Model information
axModelInfoOpenAIResponsesOpenAI: Model information
axModelInfoRekaOpenAI: Model information
axOptimizableValidatorsCommon validators reused by program-side component declarations.
axRuntimePrimitivesCanonical, ordered registry of RLM actor primitives. Order here is the order rendered into the prompt.
axSpanAttributes-
axSpanEvents-
AxStringUtil-
fFluent field builder for creating type-safe signature fields.

Functions

FunctionDescription
agent-
aiFactory function for creating AI service instances with full type safety.
axCreates a type-safe AI generator from a signature string or AxSignature object.
axAIAnthropicDefaultConfig-
axAIAnthropicVertexDefaultConfig-
axAICohereCreativeConfigCreates a creative configuration for Cohere AI service.
axAICohereDefaultConfigCreates the default configuration for Cohere AI service.
axAIDeepSeekCodeConfigCreates a configuration optimized for code generation tasks using DeepSeek V4 Pro
axAIDeepSeekDefaultConfigCreates the default configuration for DeepSeek AI with the V4 Flash model
axAIGoogleGeminiDefaultConfigAxAIGoogleGemini: Default Model options for text generation
axAIGoogleGeminiDefaultCreativeConfig-
axAIGoogleGeminiLiveAudioDefaultConfig-
axAIGrokBestConfig-
axAIGrokDefaultConfig-
axAIGrokVoiceDefaultConfig-
axAIMistralBestConfig-
axAIMistralDefaultConfig-
axAIOpenAIAudioDefaultConfig-
axAIOpenAIBestConfig-
axAIOpenAICreativeConfig-
axAIOpenAIDefaultConfig-
axAIOpenAIFastConfig-
axAIOpenAIRealtimeDefaultConfig-
axAIOpenAIRealtimeTranscriptionDefaultConfig-
axAIOpenAIResponsesBestConfig-
axAIOpenAIResponsesCreativeConfig-
axAIOpenAIResponsesDefaultConfig-
axAIRekaBestConfig-
axAIRekaCreativeConfig-
axAIRekaDefaultConfig-
axAIRekaFastConfig-
axAnalyzeChatPromptRequirementsAnalyzes a chat prompt to determine what media types it contains.
axAnalyzeRequestRequirementsAnalyzes a chat request to determine what capabilities it requires from AI providers.
axApplyOpenAIChatAudioRequest-
axAudioFormatFromMimeType-
axAudioInputFilename-
axAudioInputToBlob-
axAudioMimeType-
axBaseAIDefaultConfig-
axBaseAIDefaultCreativeConfig-
axBuildDistillerDefinitionBuilds the context-understanding actor system prompt (the distiller). The distiller explores and distills long-context inputs into a concise evidence payload for the executor stage; it has no tools.
axBuildExecutorDefinitionBuilds the executor system prompt. The executor consumes inputs.executorRequest and inputs.distilledContext from the prior distiller stage and runs tools / discovery to complete the task.
axBuildResponderDefinitionBuilds the Responder system prompt. The Responder synthesizes a final answer from the action log produced by the Actor. It NEVER generates code.
axCheckMetricsHealth-
axConcatBase64-
axCreateDefaultColorLogger-
axCreateDefaultOptimizerColorLoggerFactory function to create a default optimizer logger with color formatting
axCreateDefaultOptimizerTextLoggerFactory function to create a text-only optimizer logger (no colors)
axCreateDefaultTextLogger-
axCreateFlowColorLoggerFactory function to create a colorized AxFlow logger
axCreateFlowTextLoggerFactory function to create a text-only AxFlow logger (no colors)
axCreateGeminiLiveAudioApi-
axCreateGrokRealtimeApi-
axCreateJSRuntimeFactory function for creating an AxJSRuntime.
axCreateOpenAIRealtimeApi-
axDeserializeOptimizedProgram-
axFetchJsonSpeech-
axFetchMultipartTranscription-
axGetCompatibilityReportGenerates a comprehensive compatibility report for a request across all providers.
axGetFormatCompatibilityAnalyzes format compatibility across providers for a specific media type.
axGetMetricsConfig-
axGetOptimizerMetricsConfig-
axGetProvidersWithMediaSupportFilters providers that support a specific media type.
axGetSupportedAIModelsReturns the static Ax AI provider/model catalog.
axGoogleGeminiLiveAudioDefaults-
axIsAudioOutputEnabled-
axIsGeminiLiveAudioModel-
axIsGrokVoiceModel-
axIsOpenAIChatAudioModel-
axIsOpenAIRealtimeModel-
axIsOpenAIRealtimeTranscriptionModel-
axMapGeminiLiveAudioPart-
axMapOpenAIChatAudioDelta-
axMapOpenAIChatAudioResponse-
axMapOpenAIInputAudioPart-
axMCPToolInputSchemaToFunctionSchema-
axMergeChatAudioConfig-
axNormalizeOpenAIUsage-
axNormalizeTranscriptionResponse-
axOpenAIChatAudioDefaults-
axProcessContentForProviderProcesses content for a specific AI provider, handling unsupported media types.
axResolveGeminiLiveAudioConfig-
axResolveGrokRealtimeAudioConfig-
axResolveOpenAIChatAudioConfig-
axResolveOpenAIRealtimeAudioConfig-
axScoreProvidersForRequestScores multiple AI providers based on how well they meet request requirements.
axSelectOptimalProviderAutomatically selects the optimal AI provider for a given request.
axSerializeOptimizedProgram-
axShouldUseGeminiLiveAudio-
axShouldUseGrokRealtime-
axShouldUseOpenAIRealtime-
axUpdateMetricsConfig-
axUpdateOptimizerMetricsConfig-
axValidateChatRequestMessageValidates a chat request message item to ensure it meets the required criteria
axValidateChatResponseResultValidates a chat response result to ensure it meets the required criteria
axValidateGeminiLiveAudioInput-
axValidateProviderCapabilitiesValidates whether an AI provider can handle a request with specific requirements.
axWorkerRuntime-
bestOfN-
flow-
fnCreates a fluent builder for defining callable functions/tools with typed args, return schemas, namespaces, and optional AxAgent discovery examples.
optimize-
refine-
sCreates a type-safe signature from a string template.
Docs