Installation

Full package (recommended):

npm install orkajs

Or install only what you need:

npm install @orka-js/core @orka-js/tools @orka-js/memory

| Package | Description | |---------|-------------| | orkajs | Full package with all features | | @orka-js/core | Types, errors, utils, Knowledge | | @orka-js/openai | OpenAI adapter | | @orka-js/anthropic | Anthropic adapter | | @orka-js/mistral | Mistral adapter | | @orka-js/ollama | Ollama adapter | | @orka-js/memory | In-memory vector store | | @orka-js/pinecone | Pinecone adapter | | @orka-js/qdrant | Qdrant adapter | | @orka-js/chroma | ChromaDB adapter | | @orka-js/agent | Agents (ReAct, HITL, Toolkits) | | @orka-js/tools | Loaders, splitters, parsers, chains | | @orka-js/cache | Memory/Redis cache | | @orka-js/resilience | Retry, fallback | | @orka-js/orchestration | Router, Consensus, Race, LoadBalancer | | @orka-js/workflow | Multi-step workflows | | @orka-js/graph | Graph workflows | | @orka-js/evaluation | Metrics, test runner | | @orka-js/observability | Tracer, hooks | | @orka-js/prompts | Prompt versioning | | @orka-js/memory-store | Conversation memory |

Quick Start

import { createOrka, AnthropicAdapter, MemoryVectorAdapter } from 'orkajs';

const orka = createOrka({
  llm: new AnthropicAdapter({ apiKey: process.env.ANTHROPIC_API_KEY! }),
  vectorDB: new MemoryVectorAdapter(),
});

// Create knowledge base
await orka.knowledge.create({
  name: 'docs',
  source: ['OrkaJS is a TypeScript framework for LLM systems.'],
});

// Ask with RAG
const result = await orka.ask({
  knowledge: 'docs',
  question: 'What is OrkaJS?',
});

console.log(result.answer);

Features

| Feature | Description | |---------|-------------| | RAG | Semantic search & retrieval-augmented generation | | Agents | Structured agents with tools, policies, ReAct, Plan&Execute | | Workflows | Multi-step pipelines with plan, retrieve, generate, verify | | Adapters | Pluggable adapters for LLMs and vector databases | | Graph Workflows | Conditional branching with Mermaid export | | Multi-Model | Router, Consensus, Race, LoadBalancer orchestration | | Memory | Conversation memory with sliding window, buffer, summary | | Evaluation | Built-in metrics: relevance, faithfulness, hallucination | | Resilience | Retry, fallback, timeouts, circuit breaker patterns | | Caching | Memory & Redis cache for LLM and embeddings |

Providers

LLM: OpenAI, Anthropic, Mistral, Ollama
Vector DB: Pinecone, Qdrant, Chroma, In-Memory

Documentation

📚 orkajs.com 💬 Discord

• Getting Started
• RAG & Knowledge
• Agents
• Workflows
• Best Practices

License

MIT © Orka Team