Enables deep codebase understanding and semantic search using advanced code embeddings and knowledge graphs.
CCG-RAG (Code Guardian RAG) is a sophisticated search and retrieval engine that allows developers to interact with large repositories through semantic queries rather than simple text matching. By leveraging tree-sitter parsing for accurate code chunking and high-performance embeddings, it helps users find specific functionality, identify similar code patterns, and navigate complex documentation. It is particularly effective for onboarding to unfamiliar codebases or building context for complex refactoring tasks within the Claude Code environment, ensuring that every AI-driven change is informed by the existing architecture.
Key Features
01Knowledge graph mapping of function calls and dependency hierarchies
02Two-stage retrieval process with LLM-based reranking for high precision
03Semantic code search across the entire codebase using vector embeddings
041 GitHub stars
05Pattern-based retrieval to find similar implementations and conventions
06Intelligent documentation search for APIs, guides, and architecture specs
Use Cases
01Onboarding to a new or complex repository by exploring functionality semantically
02Finding specific logic patterns or error handlers without knowing exact text matches
03Building a comprehensive context map for large-scale code refactoring and impact analysis
