Semantic Advanced MCP Server

Advanced semantic search server with multi-format document support and reranking capabilities.

Features

• Multi-format document support:
  • PDF files
  • Word documents (DOCX)
  • Text files
• Intelligent text chunking
• Advanced semantic search with reranking
• Batch directory indexing
• Support for GPU acceleration

Installation

uv pip install -e .

Usage

Add to your Claude Desktop configuration (claude_desktop_config.json):

{
  "mcpServers": {
    "semantic-advanced": {
      "command": "uv",
      "args": [
        "--directory",
        "/opt/anaconda3/mcp-servers/semantic-advanced",
        "run",
        "python",
        "-m",
        "semantic_advanced.server"
      ]
    }
  }
}

Available Tools

1. index_directory

Index all supported documents in a directory.

Parameters:

• directory (string, required): Directory path to index
• recursive (boolean, optional): Index subdirectories recursively (default: true)

2. search

Search documents with semantic search and reranking.

Parameters:

• query (string, required): Search query
• num_results (integer, optional): Number of results to return (default: 5)
• min_similarity (float, optional): Minimum similarity threshold (default: 0.5)

Models Used

• Embeddings: BAAI/bge-base-en-v1.5
• Reranking: BAAI/bge-reranker-base

Requirements

• Python >= 3.9
• Torch
• Transformers
• ChromaDB
• FlagEmbedding