Akyn Ai

What is Akyn Ai?

Akyn Ai is an SDK that lets you create MCP (Model Context Protocol) servers from any data source. It turns documents, PDFs, websites, or raw text into a queryable knowledge base that AI assistants like Claude and Cursor can access directly — no infrastructure needed.

How to use Akyn Ai?

Install with npm install akyn-ai, then use the KnowledgeBase class to add content and serve it via stdio (for Cursor/Claude Desktop) or HTTP. You can also use the CLI with npx akyn-ai --dir ./docs --name "My Docs". Configure your MCP client as shown in the README.

Key features of Akyn Ai

• Multi-source ingestion: files, directories, URLs, raw text
• Smart text chunking with configurable size and overlap
• Flexible embeddings (OpenAI default; custom providers supported)
• In‑memory and Qdrant vector stores (extensible interface)
• Multiple transport options: stdio and HTTP
• CLI for quick indexing without writing code

Use cases of Akyn Ai

• Make project documentation searchable by Cursor or Claude Desktop
• Build Retrieval-Augmented Generation (RAG) pipelines
• Create custom AI assistants with domain‑specific knowledge
• Index research papers, guides, or any text content

FAQ from Akyn Ai

What is the difference between Akyn Ai and building an MCP server manually?

Akyn Ai provides a high‑level SDK that handles ingestion, chunking, embeddings, vector storage, and MCP transport automatically, reducing setup to a few lines of code.

What are the runtime requirements for Akyn Ai?

Node.js 18+ and an OpenAI API key (or a custom embeddings provider). The SDK is open‑source under the MIT license.

Where does the data (embeddings and documents) live?

By default, data is stored in‑memory (optionally persisted to disk). For production, you can use Qdrant (local Docker or Qdrant Cloud). You can also implement custom vector stores.

How do I connect Akyn Ai to Cursor or Claude Desktop?

Add an MCP server entry to .cursor/mcp.json or claude_desktop_config.json with the command npx ts-node ./my-kb.ts and your OpenAI API key.

How can I use Akyn Ai with custom embeddings?

Implement the EmbeddingsProvider interface (with embed and embedBatch methods) and pass it to the KnowledgeBase constructor. The default uses OpenAIEmbeddings.