grepai – semantic code search for the AI era

Summary: grepai uses vector embeddings to enable semantic search of code by functionality rather than exact text matches. It indexes code locally and supports natural language queries, call graph tracing, and real-time file watching, improving code search accuracy and efficiency for AI agents and developers.

What it does

grepai indexes code using vector embeddings to allow natural language semantic search and traces call graphs to identify function callers. It runs entirely locally and updates its index in real time by watching file changes.

Who it's for

It is designed for developers and AI agents like Claude Code and Cursor who need to find relevant code quickly without browsing numerous files or relying on exact text matches.

Why it matters

It solves the limitations of traditional grep by enabling search based on code behavior instead of names, reducing irrelevant results and token usage in AI code analysis.