FlashMLA Model Context Protocol (MCP) Server

GitHub

FlashMLA: Efficient Multi-head Latent Attention Kernels

13,180 stars C++ C++ 6 files · ~16,178 tokens

FlashMLA MCP Server

MCP server for deepseek-ai/FlashMLA

Ready-to-Use Client Configurations 
{
  "mcpServers": {
    "FlashMLA": {
      "command": "npx",
      "args": [
        "-y",
        "@deepseek-ai/FlashMLA"
      ]
    }
  }
}