grepai
Semantic Search & Call Graphs for AI Agents (100% Local)
What is grepai?
Semantic Search & Call Graphs for AI Agents (100% Local)
Overview of grepai
What it is
grepai is a privacy-first command-line interface for semantic code search. It uses vector embeddings to understand code meaning, allowing natural language queries that find relevant code even when naming conventions vary. The tool is designed to reduce AI agent input tokens by providing relevant context instead of raw search results. It runs 100% locally, meaning code never leaves the user's machine.
Key features
- Search by intent — Ask a query like “authentication logic” and find functions such as
handleUserSession. - Trace call graphs — Identify which functions call a given function before making changes.
- 100% local — All processing happens on the user's machine; no code is sent externally.
- Always up-to-date — A file watcher automatically keeps the search index fresh.
- AI agent ready — Works out of the box with Claude Code, Cursor, and Windsurf.
- MCP server — AI agents can call grepai directly as a tool via the Model Context Protocol.
Use cases
grepai is useful for developers who need to search large codebases by meaning rather than exact text patterns. It helps AI agents understand code context without sending data to external services. It can also be used to trace function call relationships before refactoring or debugging.
Getting started
Installation commands from the README:bash# Homebrew (macOS)brew install yoanbernabeu/tap/grepai``````bash# Linux/macOScurl -sSL https://raw.githubusercontent.com/yoanbernabeu/grepai/main/install.sh | sh``````powershell# Windows (PowerShell)irm https://raw.githubusercontent.com/yoanbernabeu/grepai/main/install.ps1 | iexRequires an embedding provider — Ollama (default), LM Studio, or OpenAI. For Ollama:bashollama pull nomic-embed-textQuick start commands:bashgrepai init # Initialize in your projectgrepai watch # Start indexing daemongrepai search “error handling” # Search semanticallygrepai trace callers “Login” # Find who calls a function
License and pricing
The project is open source under the MIT License, as stated in the repository metadata and README. No pricing information is provided.
Based on the project's public README and repository metadata.