File Compass

概要

File Compass revolutionizes how developers and AI assistants interact with large codebases and project directories. Instead of relying on keyword-based searches that often miss semantic context, this tool employs vector embeddings and HNSW indexing to understand the meaning behind your queries. It enables users to find specific files, functions, or concepts by simply describing what they're looking for, offering sub-second search times even across tens of thousands of files. Beyond its powerful search capabilities, File Compass integrates seamlessly with AI assistants like Claude Code through its Model Context Protocol (MCP) server, providing a suite of tools for semantic search, code preview, and quick indexing, making it an indispensable asset for AI-powered development workflows.

主な機能

• Fast Search: HNSW indexing ensures sub-second query responses for large datasets
• MCP Server: Integrates with AI assistants like Claude Code for enhanced functionality
• 0 GitHub stars
• Local Embeddings: Uses Ollama (e.g., nomic-embed-text) for privacy-preserving, local AI processing
• Semantic Search: Find files by describing intent, not just keywords
• Multi-Language AST: Semantic chunking for Python, JS, TS, Rust, Go via Tree-sitter

ユースケース

• Quickly locate specific code logic or configuration files within a large codebase by semantic description.
• Empower AI assistants to semantically search and interact with project files without needing external APIs.
• Efficiently navigate and understand new or unfamiliar repositories by describing desired functionalities.