Codebase RAG FAQs

Question 1

Can Codebase RAG manage multiple code projects?

Accepted Answer

Yes, Codebase RAG fully supports multi-project environments. It uses isolated ChromaDB collections for each distinct codebase, allowing you to index, search, and manage multiple projects independently without data mixing.

Question 2

What is Codebase RAG and what problem does it solve?

Accepted Answer

Codebase RAG is a self-hosted server providing semantic vector search over your local codebases. It solves the problem of limited codebase understanding in tools like GitHub Copilot Pro and offers a free, private alternative to paid RAG services like Cursor/Windsurf for any Model Context Protocol (MCP) client.

Question 3

Which AI clients are compatible with Codebase RAG?

Accepted Answer

Codebase RAG is designed for universal integration with any Model Context Protocol (MCP) client. This includes popular tools like GitHub Copilot (via VS Code), Cline, and Claude Desktop, allowing them to leverage your local codebase context.

Question 4

What are the core features of Codebase RAG's search capability?

Accepted Answer

Codebase RAG provides semantic vector search enhanced with keyword reranking for comprehensive results (hybrid search). It also allows configurable embedding models (e.g., sentence-transformers, Unixcoder-base) and text chunking parameters for optimized relevance.

Question 5

How does Codebase RAG keep codebases indexed and up-to-date?

Accepted Answer

Codebases are indexed automatically using vector embeddings. It offers various automation features, including watch mode for real-time file change detection, Git hooks for post-commit reindexing, and directory auto-discovery to keep your codebase search fresh and accurate.

Codebase RAG

Codebase RAG

Key Features

Use Cases

Key Features

Use Cases