OpenSearch MCP Server

MCP (Model Context Protocol) server for OpenSearch integration which is inspired by elasticsearch-mcp-server.

Features

• Index Management Tools:
  • List all indices in OpenSearch cluster
  • Get index mapping
  • Get index settings
• Cluster Management Tools:
  • Get cluster health status
  • Get cluster statistics
• Document Tools:
  • Search documents

Installation

# Clone the repository
git clone https://github.com/your-username/opensearch-mcp-server.git
cd opensearch-mcp-server

# Create and activate a virtual environment
python -m venv .venv
source .venv/bin/activate  # On Windows: .venvScriptsactivate

# Install the package
pip install uv
uv pip install -e .

Configuration

Create a .env file in the root directory with the following variables:

OPENSEARCH_HOST=https://localhost:9200
OPENSEARCH_USERNAME=xxxx
OPENSEARCH_PASSWORD=xxxx

Adjust the values to match your OpenSearch configuration.

Usage with Cursor

Run the MCP server:

uv run opensearch-mcp-server --port=<port>

Integrate with Cursor

{
  "mcpServers": {
    "opensearch": {
      "url": "http://<host>:<port>>/sse"
    }
  }
}

Usage with Claude Desktop APP

{
  "mcpServers": {
    "opensearch": {
      "command": "uv",
      "args": [
          "--directory",
          "/absolute/path/to/opensearch-mcp-server",
          "run",
          "opensearch-mcp-server"
      ]
    }
  }
}

Development

# Install dependencies
uv pip install -e .

# Run tests
uv run pytest -vv -s test_opensearch.py

License

MIT