--- title: Functions description: Grouped function reference for AI extension entry points and major subpackages deprecated: true supersededBy: ['/docs/cortex', '/docs/shield', '/docs/sentinel', '/docs/weave', '/docs/nexus'] --- ## How to Read This Page The AI module has a very large exported surface. This page is organized by the functions you typically call directly in application and extension code. ## Package `ai` (Root) ### Extension Construction and Lifecycle - `NewExtension(opts ...ConfigOption) forge.Extension` - `NewExtensionWithConfig(config Config) forge.Extension` - `(*Extension).Register(app forge.App) error` - `(*Extension).Start(ctx context.Context) error` - `(*Extension).Stop(ctx context.Context) error` - `(*Extension).Health(ctx context.Context) error` ### Config and Options - `DefaultConfig() Config` - `(Config).ToInternal(logger forge.Logger, metrics forge.Metrics) internal.AIConfig` - `WithConfig(config Config) ConfigOption` - `WithEnableLLM(enabled bool) ConfigOption` - `WithEnableAgents(enabled bool) ConfigOption` - `WithEnableTraining(enabled bool) ConfigOption` - `WithEnableInference(enabled bool) ConfigOption` - `WithEnableCoordination(enabled bool) ConfigOption` - `WithMaxConcurrency(max int) ConfigOption` - `WithRequestTimeout(timeout time.Duration) ConfigOption` - `WithCacheSize(size int) ConfigOption` - `WithLLMConfig(llm LLMConfiguration) ConfigOption` - `WithInferenceConfig(inference InferenceConfiguration) ConfigOption` - `WithAgentsConfig(agents AgentConfiguration) ConfigOption` - `WithMiddlewareConfig(middleware MiddlewareConfiguration) ConfigOption` - `WithTrainingConfig(training TrainingConfiguration) ConfigOption` - `WithStateStore(store StateStoreConfig) ConfigOption` - `WithVectorStore(store VectorStoreConfig) ConfigOption` ### Agent HTTP Controller - `NewAgentController(c forge.Container) *AgentController` - `(*AgentController).Routes(r forge.Router)` - `(*AgentController).CreateAgent(c forge.Context) error` - `(*AgentController).GetAgent(c forge.Context) error` - `(*AgentController).UpdateAgent(c forge.Context) error` - `(*AgentController).DeleteAgent(c forge.Context) error` - `(*AgentController).ListAgents(c forge.Context) error` - `(*AgentController).ExecuteAgent(c forge.Context) error` - `(*AgentController).ChatWithAgent(c forge.Context) error` - `(*AgentController).ListTemplates(c forge.Context) error` ### Agent Factory and Manager - `NewAgentFactory(llm aisdk.LLMManager, state aisdk.StateStore, logger forge.Logger, metrics forge.Metrics) *AgentFactory` - `(*AgentFactory).CreateAgent(agentType string, config map[string]any) (*aisdk.Agent, error)` - `(*AgentFactory).ListTemplates() []string` - `NewAgentManager(factory *AgentFactory, stateStore aisdk.StateStore, logger forge.Logger, metrics forge.Metrics) *AgentManager` - `(*AgentManager).CreateAgent(ctx context.Context, def *AgentDefinition) (*aisdk.Agent, error)` - `(*AgentManager).GetAgent(ctx context.Context, id string) (*aisdk.Agent, error)` - `(*AgentManager).GetDefinition(ctx context.Context, id string) (*AgentDefinition, error)` - `(*AgentManager).UpdateAgent(ctx context.Context, id string, updates *AgentDefinition) error` - `(*AgentManager).DeleteAgent(ctx context.Context, id string) error` - `(*AgentManager).ListAgents(ctx context.Context) ([]*AgentDefinition, error)` ### Store and LLM Factory Functions - `CreateStateStore(ctx context.Context, cfg StateStoreConfig, logger forge.Logger, metrics forge.Metrics) (aisdk.StateStore, error)` - `CreateVectorStore(ctx context.Context, cfg VectorStoreConfig, logger forge.Logger, metrics forge.Metrics) (aisdk.VectorStore, error)` - `CreateLLMManager(cfg LLMConfiguration, logger forge.Logger, metrics forge.Metrics) (aisdk.LLMManager, error)` ### DI Helper Functions - `GetLLMManager(c forge.Container) (aisdk.LLMManager, error)` - `MustGetLLMManager(c forge.Container) aisdk.LLMManager` - `GetAgentManager(c forge.Container) (*AgentManager, error)` - `MustGetAgentManager(c forge.Container) *AgentManager` - `GetAgentFactory(c forge.Container) (*AgentFactory, error)` - `MustGetAgentFactory(c forge.Container) *AgentFactory` - `GetStateStore(c forge.Container) (aisdk.StateStore, error)` - `MustGetStateStore(c forge.Container) aisdk.StateStore` - `GetVectorStore(c forge.Container) (aisdk.VectorStore, error)` - `MustGetVectorStore(c forge.Container) aisdk.VectorStore` - `GetModelTrainer(c forge.Container) (training.ModelTrainer, error)` - `MustGetModelTrainer(c forge.Container) training.ModelTrainer` - `GetDataManager(c forge.Container) (training.DataManager, error)` - `MustGetDataManager(c forge.Container) training.DataManager` - `GetPipelineManager(c forge.Container) (training.PipelineManager, error)` - `MustGetPipelineManager(c forge.Container) training.PipelineManager` ## Package `agents` Template constructors and lookup: - `CacheOptimizationTemplate() AgentTemplate` - `SchedulerOptimizationTemplate() AgentTemplate` - `AnomalyDetectionTemplate() AgentTemplate` - `LoadBalancerTemplate() AgentTemplate` - `SecurityMonitorTemplate() AgentTemplate` - `ResourceOptimizerTemplate() AgentTemplate` - `PredictorTemplate() AgentTemplate` - `OptimizerTemplate() AgentTemplate` - `GetTemplate(agentType string) (AgentTemplate, bool)` ## Package `inference` ### Engine Lifecycle and Inference - `NewInferenceEngine(config InferenceConfig) (*InferenceEngine, error)` - `(*InferenceEngine).Initialize(ctx context.Context, config InferenceConfig) error` - `(*InferenceEngine).Start(ctx context.Context) error` - `(*InferenceEngine).Stop(ctx context.Context) error` - `(*InferenceEngine).AddModel(model models.Model) error` - `(*InferenceEngine).RemoveModel(modelID string) error` - `(*InferenceEngine).Infer(ctx context.Context, request InferenceRequest) (InferenceResponse, error)` - `(*InferenceEngine).BatchInfer(ctx context.Context, requests []InferenceRequest) ([]InferenceResponse, error)` - `(*InferenceEngine).Scale(ctx context.Context, replicas int) error` - `(*InferenceEngine).GetStats() InferenceStats` - `(*InferenceEngine).GetHealth() InferenceHealth` ### Batching - `NewRequestBatcher(config RequestBatcherConfig) (*RequestBatcher, error)` - `(*RequestBatcher).Start(ctx context.Context) error` - `(*RequestBatcher).Stop(ctx context.Context) error` - `(*RequestBatcher).QueueRequest(ctx context.Context, request InferenceRequest) error` - `(*RequestBatcher).ProcessBatch(ctx context.Context, requests []InferenceRequest) ([]InferenceResponse, error)` - `(*RequestBatcher).QueueSize() int` - `(*RequestBatcher).GetStats() BatcherStats` ### Caching - `NewInferenceCache(config InferenceCacheConfig) (*InferenceCache, error)` - `(*InferenceCache).Start(ctx context.Context) error` - `(*InferenceCache).Stop(ctx context.Context) error` - `(*InferenceCache).Get(request InferenceRequest) (InferenceResponse, bool)` - `(*InferenceCache).Set(request InferenceRequest, response InferenceResponse)` - `(*InferenceCache).Delete(request InferenceRequest)` - `(*InferenceCache).Clear()` - `(*InferenceCache).GetStats() CacheStats` ### Pipeline and Scaling - `NewInferencePipeline(config InferencePipelineConfig) (*InferencePipeline, error)` - `(*InferencePipeline).Start(ctx context.Context) error` - `(*InferencePipeline).Stop(ctx context.Context) error` - `(*InferencePipeline).Process(ctx context.Context, request InferenceRequest) (InferenceResponse, error)` - `NewInferenceScaler(config InferenceScalerConfig) (*InferenceScaler, error)` - `(*InferenceScaler).Start(ctx context.Context) error` - `(*InferenceScaler).Stop(ctx context.Context) error` - `(*InferenceScaler).Scale(ctx context.Context, targetWorkers int) error` ## Package `training` Primary constructors and interfaces exported through `ai/exports.go`: - `training.NewModelTrainer(...)` - `training.NewDataManager(...)` - `training.NewPipelineManager(...)` - `training.ModelTrainer` interface - `training.DataManager` interface - `training.PipelineManager` interface The interface surface includes training lifecycle, dataset preparation, model save/load/export, and pipeline execution APIs. ## Package `middleware` Primary constructors: - `NewAdaptiveLoadBalance(...)` - `NewAnomalyDetection(...)` - `NewIntelligentRateLimit(...)` - `NewContentPersonalization(...)` - `NewResponseOptimization(...)` - `NewAISecurityScanner(...)` Each middleware exposes standard runtime methods: - `Name() string` - `Type() ai.AIMiddlewareType` - `Initialize(ctx context.Context, config ai.AIMiddlewareConfig) error` - `Process(ctx context.Context, req *http.Request, resp http.ResponseWriter, next http.HandlerFunc) error` - `GetStats() ai.AIMiddlewareStats` ## Package `monitoring` Primary monitoring constructors: - `NewAIHealthMonitor(...)` - `NewAIMetricsCollector(...)` - `NewAIAlertManager(...)` - `NewAIDashboard(...)` These provide independent health checks, metrics aggregation, alerting, and dashboard APIs. ## Source Navigation Tip For full symbol exploration across all subpackages: ```bash go doc github.com/xraph/forge/extensions/ai/... ``` Use this together with this page to separate extension entry points from advanced package internals.