gpu-search-mcp

Alexlexer/gpu-search-mcp
★ 0 stars Python 📁 File System Updated 5d ago
GPU-accelerated codebase search MCP server — loads source files into RTX VRAM, searches with PyTorch CUDA ops
View on GitHub →

Quick Install

Copy the config for your editor. Some servers may need additional setup — check the README.

Add to claude_desktop_config.json:

{
  "mcpServers": {
    "gpu-search-mcp": {
      "command": "uvx",
      "args": [
        "gpu-search-mcp"
      ]
    }
  }
}

Or install with pip: pip install gpu-search-mcp

README Excerpt

A GPU-accelerated codebase search server built as an [MCP](https://modelcontextprotocol.io/) tool. It loads your source files directly into RTX VRAM and runs searches as vectorized CUDA operations via PyTorch — no custom kernels, no native extensions. > **Status:** Working prototype, used daily on a single machine. Core search is solid; some features described below are best-effort (see [Limitations](#known-limitations)).

Tools (8)

ICodeEditoraddEventListenercontextModecreateTextModeldisposeOnReturnhandleErrormodetopK