AgenticRAG

소개

AgenticRAG is an intelligent codebase processing server offering agentic Retrieval-Augmented Generation (RAG) capabilities through the Model Context Protocol (MCP). It automatically indexes, chunks, and embeds codebases, enabling semantic search and providing comprehensive context through a self-critiquing retrieval loop. With multi-model architecture using GPT-4o and Claude 3, live updates via file system watching, and built-in cost control, it offers an efficient solution for understanding and interacting with code.

주요 기능

• Agentic Retrieval: Self-critiquing retrieval loop that ensures comprehensive context
• Live Updates: File system watching for automatic re-indexing
• 0 GitHub stars
• Cost Control: Built-in telemetry and budget management
• Intelligent Code Indexing: Automatically chunks and embeds codebases for semantic search
• Multi-Model Architecture: Uses GPT-4o for retrieval and Claude 3 for planning

사용 사례

• Understand complex code structures and dependencies using natural language queries.
• Index and search large codebases to quickly find relevant information.
• Get statistics on a project's code, such as the number of files, chunks, and tokens.