zvec-grep (zg): A Local-First Hybrid Search Engine for Humans and AI Agents

For years, ripgrep has been the go-to tool for developers seeking to search through massive codebases swiftly, offering instant multi-threaded regex matching. However, traditional grep tools suffer from a significant limitation—they necessitate users to know the exact string literals, variable names, or regex patterns ahead of time.

This requirement leads to cumbersome guessing processes when developers or AI coding agents explore unfamiliar codebases, attempting to locate components or understand system architecture. Enter zvec-grep (zg), an open-source search engine created by zvec-ai, harnessing Alibaba's high-performance zvec vector library. zg unifies the strengths of ripgrep, BM25, and local vector search into a single command-line tool, seamlessly integrating with AI agent environments. zg stands out as a local-first search layer, catering symmetrically to both human developers and autonomous AI coding agents.

It indexes source code, documents, and configuration files locally, facilitating queries that blend the exactness of keyword matching with the discovery capabilities of semantic embeddings. zg's key features include Hybrid Search (Semantic + BM25 + Ripgrep), Agent-Native Context Optimization, 100% Local & Privacy-Preserving capabilities, and Dual CLI and MCP Integration.

By delivering lightning-fast results while preserving user privacy, zg revolutionizes the way developers and AI agents navigate and interact with code repositories.