NeuroLink — The Nervous System for AI Streams

# NeuroLink SDK > NeuroLink is an open-source TypeScript AI SDK providing unified access to 13+ AI providers (OpenAI, Anthropic, Google AI Studio, Google Vertex AI, AWS Bedrock, Azure OpenAI, Mistral, LiteLLM, Ollama, Hugging Face, SageMaker, OpenRouter, and OpenAI-compatible endpoints) through a single consistent API. Ships as TypeScript SDK + CLI with comprehensive multimodal support (50+ file types), RAG (9 chunking strategies, 22 vector stores, hybrid search), MCP integration (58+ servers, 4 transports), multi-agent workflows with 3 topologies, 3-layer memory system, voice processing (8 TTS + 6 STT providers), workflow engine with checkpointing, and enterprise observability. Extracted from production systems at Juspay. MIT licensed. ## Getting Started - [Installation & Quick Start](https://docs.neurolink.ink/docs/getting-started): Install via npm and make your first API call in 4 lines of TypeScript - [Provider Setup](https://docs.neurolink.ink/docs/providers): Configure any of 13+ supported AI providers ## Core Features - [RAG Processing](https://docs.neurolink.ink/docs/rag/overview): 9 chunking strategies, 22 vector stores, hybrid search (BM25 + vector), 5 rerankers - [MCP Integration](https://docs.neurolink.ink/docs/mcp/overview): 58+ external servers, 4 transport types (stdio, HTTP, SSE, WebSocket), ToolRouter, ToolCache, RequestBatcher - [Multi-Agent Systems](https://docs.neurolink.ink/docs/agents/overview): 3 topologies (hub-spoke, mesh, pipeline), AgentNetwork orchestration - [Streaming](https://docs.neurolink.ink/docs/advanced/streaming): 24 event types, backpressure control, real-time processing - [Voice](https://docs.neurolink.ink/docs/features/voice): 8 TTS providers, 6 STT providers, speech-to-speech pipeline - [Multimodal](https://docs.neurolink.ink/docs/features/multimodal): 50+ file types — images, PDFs, CSV, Excel, Word, code, video, audio, archives - [Memory](https://docs.neurolink.ink/docs/features/memory): 3-layer architecture — conve