Chunker
Createdtarjeir
Chunks source code files and stores them in a ChromaDB vector database, including line range metadata.
About
Chunker is a command-line tool and Model Context Protocol (MCP) server designed to improve code understanding and analysis. It leverages LangChain's RecursiveCharacterTextSplitter to divide source code into manageable chunks and stores these chunks in a ChromaDB vector database. Each chunk is enriched with metadata, including the file path and line range, facilitating precise and context-aware querying. This enables developers to quickly find relevant code snippets, understand code structure, and integrate code analysis into workflows like Claude for Desktop.
Key Features
- Chunks code files using LangChain's RecursiveCharacterTextSplitter.
- Stores chunks in ChromaDB with path and line range metadata.
- Supports command-line interface using Typer.
- Asynchronous, batched insertion into ChromaDB for performance.
- Provides an MCP server endpoint for integrations with tools like Claude for Desktop.
- 0 GitHub stars
Use Cases
- Integrating with Claude for Desktop to enable code chunking and vectorisation directly from Claude's interface.
- Querying ChromaDB collection for relevant code chunks using CLI.
- Analyzing code structure by leveraging file path and line range metadata.
