VectorCode MCP

VectorCode exposes local repository indexing and semantic code retrieval through a Python MCP server with explicit project roots.

Local serverstdioPythonMIT

What is the VectorCode MCP Server?

VectorCode exposes local repository indexing and semantic code retrieval through a Python MCP server with explicit project roots.

Retrieve code by meaning

VectorCode vectorizes repository files and uses ChromaDB and embedding models to retrieve relevant code. The MCP interface includes collection inspection, indexing, queries and indexed-file management. It can support an assistant that needs related implementations before proposing a change.

Install the MCP extra

The Python project defines a vectorcode-mcp-server entry point and an mcp optional dependency group. Keep a compatible SDK version and Python environment, then register the local stdio server. The source metadata requires Python 3.11 through 3.13. A project root supplied to the tools identifies the repository being indexed or queried.

Check retrieval against source files

Index a small repository and query a function whose location is known. Compare returned passages with the current files, including their paths and scope. Re-index changed files before depending on an older vector representation. Removing a file from the index is different from deleting the source file; inspect the documented operation and returned result when managing indexed content.

Project documentation

Repository and setup guide

How to install the VectorCode MCP Server

{
  "mcpServers": {
    "vectorcode": {
      "command": "vectorcode-mcp-server",
      "args": []
    }
  }
}

Configuration

Python 3.11–3.13, VectorCode with the MCP extra, embedding resources and the intended repository.

Frequently asked questions

The Python project defines an mcp extra and a vectorcode-mcp-server console entry point.