mcp -opensearch

This MCP server provides a semantic memory layer on top of an OpenSearch cluster. It uses OpenSearch as the backing store to store and retrieve memories via MCP tools. The primary tool exposed is search-openSearch, which accepts a structured query payload and returns a confirmation or results that can be fed into LLM workflows. You can run the server via uvx for Python-based setups, then connect your MCP-enabled application or Claude Desktop to perform memory storage and lookup operations. The examples in the repository show how to configure Claude Desktop to launch the server and how to test the OpenSearch connection locally. Once running, you can issue memory storage operations and semantic queries that leverage OpenSearch’s search capabilities through the MCP protocol.

Prerequisites:

• Python and uv (or uvx) installed on your system
• Access to an OpenSearch cluster (locally at http://localhost:9200 by default) or a reachable OpenSearch endpoint

Installation steps:

1. Install Smithery (optional for automated installation):

# If you want to install via Smithery (optional)
npx -y @smithery/cli install @ibrooksSDX/mcp-server-opensearch --client claude

2. Install and run the MCP server using uv (recommended):

# Start the MCP server with OpenSearch configuration
uv run mcp-server-opensearch \
  --opensearch-url "http://localhost:9200" \
  --index-name "your_index_name"

3. If you prefer Claude Desktop integration, add the opensearch server to claude_desktop_config.json as shown in the repository README, or use the FastMCP installer UI to load the server.
4. Optional: Verify connectivity with a local test:

uv run python src/mcp-server-opensearch/test_opensearch.py

Tips and caveats:

• Ensure your OpenSearch instance is reachable at the configured URL and that the index exists or can be auto-created by the client.
• The OpenSearch async client may require extra dependencies; if installation fails, try the standard synchronous client or verify environment compatibility.
• Environment variable mappings (OPENSEARCH_HOST, OPENSEARCH_HOSTPORT, INDEX_NAME) can be used to parameterize deployments in containerized or cloud environments.
• If you encounter authentication errors, provide the appropriate API key or credentials required by your OpenSearch deployment.
• The mcpServers configuration shown in Claude Desktop examples is a reference; adjust paths and flags to match your local setup and workflow.