DevRag

About

DevRag is a lightweight, cross-platform RAG system specifically engineered for developers utilizing Claude Code. It revolutionizes how developers interact with their local markdown documentation by replacing costly full document reads with precise vector search. By indexing markdown files locally using multilingual embeddings, DevRag enables natural language queries, drastically reducing token consumption (up to 40x less) and search times (up to 15x faster) in Claude Code. It's a single binary solution that automatically downloads models and supports incremental syncing, making local documentation instantly discoverable and efficient for both individual and team development workflows.

Key Features

• Simple RAG for Claude Code
• Markdown Support with auto-indexing
• Semantic Search using natural language queries
• Single Binary with no Python dependency, models auto-download
• Multilingual support for 100+ languages
• 13 GitHub stars

Use Cases

• Saving tokens and accelerating information retrieval when using Claude Code with local documentation.
• Efficiently searching large personal or project documentation in markdown format using natural language.
• Enabling team members to collaboratively search shared documentation repositories managed in Git with local, auto-syncing indexes.