PyVector

소개

PyVector offers a robust yet minimal vector database solution tailored for local Large Language Model (LLM) development. It distinguishes itself by providing an MCP server for seamless LLM tool integration and an HTTP server for broader application compatibility, all while maintaining minimal dependencies. Designed with flexibility in mind, PyVector includes intelligent fallback implementations for both vector indexing (NumPy instead of FAISS) and embedding generation (hash-based instead of sentence-transformers), ensuring functionality even in resource-constrained environments while enabling performance upgrades when optional libraries are present.

주요 기능

• Text embedding generation using sentence transformers or hash-based fallback
• MCP server for seamless LLM tool use
• Minimal dependencies (NumPy and Pydantic)
• Fast similarity search with FAISS or NumPy fallback
• 0 GitHub stars
• HTTP server for REST API integration

사용 사례

• Integrating vector search capabilities into LLM tools via the MCP
• Developing local LLM applications requiring semantic search
• Providing similarity search through a local HTTP API