MCPunk FAQs

Question 1

What is MCPunk?

Accepted Answer

MCPunk enables you to chat with your codebase by providing an LLM with tools to intelligently search your code. It breaks files into chunks and lets the LLM find the code needed to answer your questions.

Question 2

How does MCPunk work without embeddings?

Accepted Answer

MCPunk uses a "roaming RAG" approach. It chunks code into logical units and provides the LLM tools to search and query these chunks directly, eliminating the need for vector embeddings.

Question 3

What file types does MCPunk support?

Accepted Answer

MCPunk includes built-in chunkers for Python, Vue, and Markdown files. It also has a fallback chunker that handles any other file type by treating the entire file as a single chunk.

Question 4

Can I see what code MCPunk is using to answer my questions?

Accepted Answer

Yes! MCPunk provides clear, auditable search results, allowing you to see exactly what data the LLM has considered and how it found it. This transparency makes it easier to guide the process.

Question 5

What is 'roaming RAG'?

Accepted Answer

Roaming RAG is a method where an LLM is given tools to navigate and search through structured chunks of content (like code).  Instead of vector search, the LLM actively explores and finds relevant information.

MCPunk

About

Key Features

Use Cases