Semantic Code Search MCP

Example Use Cases

• 1

  Perform semantic search over codebases to find relevant code snippets by meaning rather than text matching.

• 2

  Incremental and efficient indexing of large code projects with support for Python, Rust, and Markdown.

• 3

  That require integration with Claude Code for enhanced code understanding and retrieval.

Description

This MCP server indexes codebases by parsing source files into meaningful chunks, embedding them with sentence-transformers, and storing vectors in LanceDB for efficient semantic search. It supports incremental indexing and hybrid semantic plus keyword search with recency boosting, enabling more relevant and meaningful code search results compared to traditional text matching. The server integrates with Claude Code and can be extended to support additional programming languages via tree-sitter chunkers.