JavaScript API Reference¶

Complete reference for GraphBit's JavaScript API. This document covers all classes, methods, and their usage based on the actual JavaScript binding implementation.

Module: @infinitibit_gmbh/graphbit¶

Core Functions¶

init(options?)¶

Initialize the GraphBit library with optional configuration.

import { init } from '@infinitibit_gmbh/graphbit';

// Basic initialization
init();

// With options
init({
  logLevel: 'info',
  enableTracing: true,
  debug: true
});

Parameters: - options (object, optional): Configuration options - logLevel (string): Log level ("trace", "debug", "info", "warn", "error"). Default: "warn" - enableTracing (boolean): Enable tracing. Default: false - debug (boolean): Enable debug mode. Default: false

Returns: void
Throws: Error if initialization fails

version()¶

Get the current GraphBit version.

import { version } from '@infinitibit_gmbh/graphbit';

const v = version();
console.log(`GraphBit version: ${v}`);

Returns: string - Version string (e.g., "0.5.1")

getSystemInfo()¶

Get comprehensive system information and health status.

import { getSystemInfo } from '@infinitibit_gmbh/graphbit';

const info = getSystemInfo();
console.log(`Node.js version: ${info.nodeVersion}`);
console.log(`CPU count: ${info.cpuCount}`);

Returns: object - Dictionary containing: - version: GraphBit version - nodeVersion: Node.js version - runtimeUptime: Runtime uptime in milliseconds - cpuCount: Number of CPU cores - buildTarget: Build target platform - buildProfile: Build profile (debug/release)

healthCheck()¶

Perform comprehensive health checks.

import { healthCheck } from '@infinitibit_gmbh/graphbit';

const health = healthCheck();
if (health.overallHealthy) {
  console.log('System is healthy');
} else {
  console.log('System has issues');
}

Returns: object - Object containing health status information

configureRuntime(config)¶

Configure runtime parameters.

import { configureRuntime } from '@infinitibit_gmbh/graphbit';

configureRuntime({
  maxThreads: 4,
  enableMonitoring: true
});

Parameters: - config: Runtime configuration object

LLM Configuration¶

LlmConfig¶

Configuration class for Large Language Model providers.

Static Methods¶

LlmConfig.openai(options)¶

Create OpenAI provider configuration.

import { LlmConfig } from '@infinitibit_gmbh/graphbit';

const config = LlmConfig.openai({
  apiKey: process.env.OPENAI_API_KEY,
  model: 'gpt-4o-mini'
});

LlmConfig.anthropic(options)¶

Create Anthropic provider configuration.

const config = LlmConfig.anthropic({
  apiKey: process.env.ANTHROPIC_API_KEY,
  model: 'claude-3-5-sonnet-20241022'
});

LlmConfig.ollama(options)¶

Create Ollama (local) provider configuration.

const config = LlmConfig.ollama({
  baseUrl: 'http://localhost:11434',
  model: 'llama2'
});

LlmConfig.azureOpenai(options)¶

Create Azure OpenAI provider configuration.

const config = LlmConfig.azureOpenai({
  apiKey: process.env.AZURE_API_KEY,
  endpoint: 'https://...',
  deploymentName: 'gpt-4'
});

LlmConfig.deepseek(options)¶

Create DeepSeek provider configuration.

const config = LlmConfig.deepseek({
  apiKey: process.env.DEEPSEEK_API_KEY,
  model: 'deepseek-chat'
});

Other Providers¶

Supported methods: bytedance, huggingface, perplexity, openrouter, fireworks, replicate, togetherai, xai, ai21, mistralai.

LLM Client¶

LlmClient¶

Direct language model access without workflow overhead.

Constructor¶

const client = new LlmClient(config);

Methods¶

complete(prompt, maxTokens)¶

Simple text completion.

const text = await client.complete('Hello world');

Returns: Promise<string>

completeFull(prompt, maxTokens)¶

Full response with metadata.

const response = await client.completeFull('Hello world');
console.log(response.content, response.usage);

Returns: Promise<LlmResponse>

Workflow¶

WorkflowBuilder¶

Builder for creating workflows.

const builder = new WorkflowBuilder('My Workflow');
builder.description('Processes data');
const workflow = builder.build();

Workflow¶

Workflow representation.

Methods¶

addNode(node)¶

Add execution node to workflow.

await workflow.addNode({
  id: 'node1',
  nodeType: 'Agent',
  // ... configuration
});

validate()¶

Validate workflow structure.

await workflow.validate();

Executor¶

Executor¶

Workflow execution engine.

Constructor¶

const executor = new Executor(llmConfig, {
  timeoutSeconds: 300,
  maxParallel: 4
});

Methods¶

execute(workflow)¶

Execute a workflow.

const context = await executor.execute(workflow);

Returns: Promise<WorkflowContext>

WorkflowContext¶

Execution results and verification.

Methods¶

• isCompleted(): Promise<boolean>
• isFailed(): Promise<boolean>
• error(): Promise<string | null>
• getAllOutputs(): Promise<string>
• stats(): Promise<WorkflowExecutionStats | null>

Agents¶

AgentBuilder¶

Builder for creating agents.

const agent = new AgentBuilder('My Agent', llmConfig)
  .description('Helper agent')
  .systemPrompt('You are helpful')
  .build();

Document Processing¶

DocumentLoader¶

Methods¶

loadFile(path, type)¶

Load a document from a file.

const doc = await loader.loadFile('doc.pdf', 'pdf');

loadText(text, source)¶

Load from a string.

const doc = await loader.loadText('content', 'source_id');

Static Methods¶

DocumentLoader.supportedTypes()¶

Get list of supported extensions.

DocumentLoader.detectDocumentType(path)¶

Detect type from file path.

Embeddings¶

EmbeddingConfig¶

Configuration for embeddings (e.g., EmbeddingConfig.openai(...)).

EmbeddingClient¶

Methods¶

embed(texts)¶

Generate embeddings for an array of strings.

const result = await client.embed(['text1', 'text2']);

EmbeddingClient.similarity(embedding1, embedding2)¶

Calculate cosine similarity (static method).

const score = EmbeddingClient.similarity(emb1, emb2);

Tool Registry¶

ToolRegistry¶

Methods¶

• register(name, description, schema, callback)
• execute(name, args)
• getToolCount()
• getLlmTools(): Get tools formatted for LLM usage.