elsium-ai

Single import for the entire ElsiumAI framework.

Install

npm install elsium-ai

Why use this package?

elsium-ai is the umbrella package that re-exports every public API from all ElsiumAI sub-packages. Instead of installing and importing from many individual packages, you can import everything from a single entry point.

Before -- importing from individual sub-packages:

import { env } from '@elsium-ai/core'
import { gateway } from '@elsium-ai/gateway'
import { defineAgent } from '@elsium-ai/agents'

After -- importing from the umbrella package:

import { env, gateway, defineAgent } from 'elsium-ai'

Same APIs, one dependency, one import path.

What's Included

Every value and type listed below is available directly from 'elsium-ai'.

@elsium-ai/core

| Values | Types | |--------|-------| | ElsiumError | Role | | ok | ContentPart | | err | TextContent | | isOk | ImageContent | | isErr | AudioContent | | unwrap | DocumentContent | | unwrapOr | Message | | tryCatch | ToolCall | | tryCatchSync | ToolResult | | createStream | TokenUsage | | createLogger | CostBreakdown | | env | StopReason | | envNumber | LLMResponse | | envBool | StreamEvent | | generateId | XRayData | | generateTraceId | StreamCheckpoint | | extractText | ProviderConfig | | sleep | CompletionRequest | | retry | ToolDefinition | | zodToJsonSchema | TenantContext | | createRegistry | Middleware | | countTokens | Result | | createContextManager | Ok | | | Err | | | ElsiumStream | | | ResilientStreamOptions | | | LogLevel | | | Logger | | | ErrorCode | | | Registry | | | ContextStrategy | | | ContextManagerConfig | | | StreamMiddleware | | | StreamMiddlewareNext | | | ContextManager |

@elsium-ai/gateway

| Values | Types | |--------|-------| | gateway | LLMProvider | | registerProviderFactory | ProviderFactory | | calculateCost | Gateway | | estimateCost | | | composeStreamMiddleware | | | registerPricing | GatewayConfig | | composeMiddleware | XRayStore | | loggingMiddleware | ProviderMeshConfig | | costTrackingMiddleware | ProviderEntry | | xrayMiddleware | RoutingStrategy | | createAnthropicProvider | ProviderMesh | | createOpenAIProvider | SecurityMiddlewareConfig | | createGoogleProvider | SecurityViolation | | createProviderMesh | SecurityResult | | securityMiddleware | CacheAdapter | | detectPromptInjection | CacheStats | | detectJailbreak | CacheMiddlewareConfig | | redactSecrets | OutputGuardrailConfig | | checkBlockedPatterns | OutputGuardrailRule | | cacheMiddleware | OutputViolation | | createInMemoryCache | BatchConfig | | outputGuardrailMiddleware | BatchResult | | createBatch | BatchResultItem |

@elsium-ai/agents

| Values | Types | |--------|-------| | defineAgent | Agent | | runSequential | AgentDependencies | | runParallel | AgentConfig | | runSupervisor | AgentResult | | createMemory | AgentRunOptions | | createSharedMemory | SharedMemory | | createSemanticValidator | MultiAgentOptions | | createAgentSecurity | GuardrailConfig | | createConfidenceScorer | AgentHooks | | executeStateMachine | Memory | | createInMemoryMemoryStore | MemoryConfig | | createSqliteMemoryStore | MemoryStore | | | SqliteMemoryStoreConfig | | | StateTransitionResult | | | SemanticGuardrailConfig | | | SemanticCheck | | | SemanticCheckResult | | | SemanticValidationResult | | | SemanticValidator | | | AgentSecurityConfig | | | AgentSecurityResult | | | ConfidenceConfig | | | ConfidenceResult | | | StateDefinition | | | StateHistoryEntry | | | StateMachineResult |

@elsium-ai/tools

| Values | Types | |--------|-------| | defineTool | Tool | | createToolkit | ToolConfig | | httpFetchTool | ToolContext | | calculatorTool | ToolExecutionResult | | jsonParseTool | Toolkit | | currentTimeTool | | | formatToolResult | | | formatToolResultAsText | |

@elsium-ai/rag

| Values | Types | |--------|-------| | rag | RAGPipeline | | createInMemoryStore | RAGPipelineConfig | | createOpenAIEmbeddings | IngestResult | | createGoogleEmbeddings | Document | | createCohereEmbeddings | Chunk | | createMockEmbeddings | EmbeddedChunk | | vectorStoreRegistry | RetrievalResult | | embeddingProviderRegistry | QueryOptions | | createPgVectorStore | EmbeddingProvider | | createQdrantStore | VectorStore | | createBM25Index | VectorStoreFactory | | createHybridSearch | EmbeddingProviderFactory | | | PgVectorStoreConfig |

@elsium-ai/workflows

| Values | Types | |--------|-------| | defineWorkflow | Workflow | | defineParallelWorkflow | WorkflowConfig | | defineBranchWorkflow | WorkflowResult | | defineDagWorkflow | WorkflowRunOptions | | step | StepConfig | | | StepContext | | | StepResult | | | DagStepConfig | | | DagWorkflowConfig |

@elsium-ai/observe

| Values | Types | |--------|-------| | observe | Tracer | | createSpan | TracerConfig | | createMetrics | TracerExporter | | createCostEngine | CostReport | | createExperiment | Span | | instrumentComplete | ExperimentStore | | instrumentAgent | | | createFileExperimentStore | | | toOTelSpan | SpanData | | toOTelExportRequest | SpanKind | | toTraceparent | SpanStatus | | parseTraceparent | MetricsCollector | | injectTraceContext | MetricEntry | | extractTraceContext | CostEngine | | createOTLPExporter | CostEngineConfig | | | CostAlert | | | CostDimension | | | CostIntelligenceReport | | | ModelSuggestion | | | Experiment | | | ExperimentConfig | | | ExperimentVariant | | | ExperimentResults | | | OTelSpan | | | OTelExportRequest | | | TraceContext | | | OTLPExporterConfig |

@elsium-ai/app

| Values | Types | |--------|-------| | createApp | AppConfig | | sseHeaders | ServerConfig | | formatSSE | CorsConfig | | streamResponse | AuthConfig | | tenantMiddleware | RateLimitConfig | | tenantRateLimitMiddleware | StreamChatEvent | | tenantBudgetMiddleware | StreamCompleteEvent | | | TenantMiddlewareConfig |

@elsium-ai/mcp

| Values | Types | |--------|-------| | createMCPClient | MCPClient | | createMCPServer | MCPClientConfig | | createMCPHttpHandler | MCPToolInfo | | | MCPServer | | | MCPServerConfig | | | JsonRpcRequest | | | JsonRpcResponse | | | MCPTransport |

@elsium-ai/client

| Values | Types | |--------|-------| | createClient | ElsiumClient | | | ClientConfig |

@elsium-ai/testing

| Values | Types | |--------|-------| | mockProvider | MockProviderOptions | | createFixture | MockResponseConfig | | loadFixture | MockProvider | | createRecorder | EvalSuiteConfig | | runEvalSuite | EvalCase | | formatEvalReport | EvalCriterion | | createSnapshotStore | EvalResult | | createPromptRegistry | EvalSuiteResult | | definePrompt | LLMJudge | | createRegressionSuite | SnapshotStore | | createReplayRecorder | PromptDefinition | | createReplayPlayer | PromptDiff | | | PromptRegistry | | | RegressionBaseline | | | RegressionResult | | | RegressionDetail | | | RegressionSuite | | | ReplayEntry | | | ReplayRecorder | | | ReplayPlayer |

Quick Example

import {
  env,
  gateway,
  defineAgent,
  defineTool,
  observe,
  createCostEngine,
} from 'elsium-ai'
import { z } from 'zod'

// Set up observability
const tracer = observe({ serviceName: 'my-app' })
const costEngine = createCostEngine({ budget: { daily: 10.0 } })

// Create a gateway to an LLM provider
const llm = gateway({
  provider: 'anthropic',
  model: 'claude-sonnet-4-6',
  apiKey: env('ANTHROPIC_API_KEY'),
})

// Define a tool
const weatherTool = defineTool({
  name: 'get_weather',
  description: 'Get current weather for a city',
  parameters: z.object({ city: z.string() }),
  execute: async ({ city }) => ({ temp: 72, city }),
})

// Define an agent with the tool
const agent = defineAgent(
  {
    name: 'assistant',
    system: 'You are a helpful assistant.',
    tools: [weatherTool],
  },
  { complete: (req) => llm.complete(req) },
)

// Run the agent
const result = await agent.run('What is the weather in Paris?')
console.log(result.output)

Individual Packages

For full API documentation, see each sub-package README.

| Package | npm | Description | |---------|-----|-------------| | @elsium-ai/core | npm | Types, errors, result pattern, streaming, logger, config, tokens, context manager, registry, schema | | @elsium-ai/gateway | npm | Multi-provider LLM gateway, middleware, provider mesh, security, caching, output guardrails, batch | | @elsium-ai/agents | npm | Agent definitions, memory, persistent stores (in-memory, SQLite), guardrails, multi-agent | | @elsium-ai/tools | npm | Tool definitions with Zod validation, built-in tools | | @elsium-ai/rag | npm | Document ingestion, chunking, embeddings, vector search, PgVector store, plugin registries | | @elsium-ai/workflows | npm | Sequential, parallel, and branching workflow definitions | | @elsium-ai/observe | npm | Tracing, metrics, cost intelligence, OpenTelemetry export, A/B experiments | | @elsium-ai/app | npm | HTTP server with CORS, auth, rate limiting, SSE streaming, multi-tenant | | @elsium-ai/client | npm | TypeScript HTTP client with SSE parsing for consuming ElsiumAI servers | | @elsium-ai/mcp | npm | Model Context Protocol client and server | | @elsium-ai/testing | npm | Mocks, fixtures, evals, snapshot testing, prompt versioning, replay |

License

MIT - Copyright (c) 2026 Eric Utrera