File Compass

关于

File Compass offers an intelligent way for AI workstations to navigate large codebases. Instead of relying on rigid keyword searches, it leverages semantic search powered by local embeddings via Ollama and efficient HNSW vector indexing. This allows users to describe what they are looking for in natural language, enabling rapid discovery of files, functions, or sections of code based on their meaning, even across diverse programming languages. It integrates seamlessly with AI assistants like Claude Code via an MCP server, providing a suite of file interaction tools.

主要功能

• 0 GitHub stars
• Semantic Search for descriptive queries
• Multi-Language AST-aware Chunking (Python, JS, TS, Rust, Go)
• MCP Server for AI assistant integration (e.g., Claude Code)
• Local Embeddings via Ollama (no API keys required)
• Fast HNSW Indexing for sub-second query latency

使用案例

• Quickly locate specific code files or sections within large projects based on their functionality or purpose.
• Integrate advanced file search capabilities into AI assistants for enhanced code understanding and interaction.
• Efficiently navigate and understand new or unfamiliar codebases by describing desired content.