Show HN: cgrep – local, code-aware search for AI coding agents (github.com)

The recent launch of cgrep, a local, code-aware search tool for AI coding agents, marks a significant advancement in code retrieval efficiency. Currently at version 1.4.1, cgrep integrates innovative features such as BM25 full-text search, AST symbol extraction, and optional semantic search with embeddings, all designed to help both humans and AI agents navigate real codebases. Notably, in benchmarking scenarios involving PyTorch, cgrep demonstrated a remarkable reduction of tokens-to-complete by 95.2% and improved average retrieval latency by about 58.2 times compared to traditional grep. This tool emphasizes a local-first approach, enhancing both speed and privacy while enabling scalable workflows on large repositories. With ergonomic command-line shortcuts, deterministic JSON output for seamless agent interactions, and built-in indexing capabilities, cgrep is set to streamline workflows for developers and AI agents alike. By transforming how code intent is searched and interpreted, cgrep may significantly change the landscape of AI-assisted coding and contribute to more effective development practices in the AI and machine learning communities.