Local Code Indexing for Cursor

What is Local Code Indexing for Cursor?

An experimental Python-based server that locally indexes codebases using ChromaDB and provides a semantic search tool via an MCP (Model Context Protocol) server for tools like Cursor.

How to use Local Code Indexing for Cursor?

Clone the repository, create a .env file with PROJECTS_ROOT and FOLDERS_TO_INDEX, then start the server with docker-compose up -d. Configure Cursor by adding the server URL http://localhost:8978/sse to ~/.cursor/mcp.json. Optionally create a .cursorrules file to prompt the Cursor Agent to use the @search_code tool.

Key features of Local Code Indexing for Cursor

• Fully local code indexing using ChromaDB
• Semantic code search via MCP for Cursor
• Configurable project root and folders to index
• Docker-based deployment for easy setup
• SSE transport on port 8978

Use cases of Local Code Indexing for Cursor

• Enable Cursor Agent to semantically search your codebase instead of relying on grep
• Quickly find relevant code across multiple projects without leaving the IDE
• Offline, privacy-preserving code search that never sends code to external servers

FAQ from Local Code Indexing for Cursor

What is Local Code Indexing for Cursor?

It is an experimental Python server that indexes your local projects using ChromaDB and exposes a semantic search tool via MCP for use in Cursor.

How do I configure which folders are indexed?

Set PROJECTS_ROOT and FOLDERS_TO_INDEX in your .env file before starting the server.

Does it require Docker?

Yes, the recommended way to run the server is using Docker Compose (docker-compose up -d).

How do I connect it to Cursor?

Edit ~/.cursor/mcp.json to include the server URL http://localhost:8978/sse under mcpServers.workspace-code-search, then restart Cursor.

Where does the indexed data live?

The README does not specify the storage location, but since it uses ChromaDB and runs locally, data is stored on your machine.