Search through a specified Trieve dataset with support for semantic, fulltext, hybrid, or BM25 search types
Trieve MCP Server provides AI agents with tools to search and interact with Trieve datasets. It enables semantic, fulltext, hybrid, and BM25 search capabilities across your organization's datasets through a standardized interface. With Trieve MCP Server, AI assistants can access and search through your knowledge base, making it easier to retrieve relevant information during conversations. The server supports advanced filtering options and customizable result highlighting to ensure you get the most relevant search results.
Trieve MCP Server enables AI agents to search and interact with your Trieve datasets through a standardized interface. This integration allows AI assistants like Claude to search through your organization's knowledge base and retrieve relevant information during conversations.
The easiest way to install Trieve MCP Server for Claude Desktop is via Smithery:
npx -y @smithery/cli install trieve-mcp-server --client claude
You can also install the package manually:
npm install trieve-mcp-server
Trieve MCP Server requires authentication with your Trieve account. You'll need to provide your API key and organization ID, which you can find in your dashboard at dashboard.trieve.ai.
You can configure the server using environment variables:
TRIEVE_API_KEY (required): Your Trieve API keyTRIEVE_ORGANIZATION_ID (required): Your Trieve organization IDTRIEVE_DATASET_ID (optional): Specific dataset ID to useAlternatively, you can provide configuration via command-line arguments, which will override any environment variables:
trieve-mcp-server --api-key <your-api-key> --org-id <your-org-id> [--dataset-id <dataset-id>]
To use Trieve with Claude Desktop, add the following configuration to your claude_desktop_config.json file:
{
"mcpServers": {
"trieve-mcp-server": {
"command": "npx",
"args": ["trieve-mcp-server@latest"],
"env": {
"TRIEVE_API_KEY": "your-api-key-here",
"TRIEVE_ORGANIZATION_ID": "your-org-id-here",
"TRIEVE_DATASET_ID": "optional-dataset-id-here"
}
}
}
}
Replace the placeholder values with your actual Trieve credentials.
Once configured, Claude Desktop will automatically detect available Trieve datasets as attachments. You can select a dataset and then begin chatting with Claude about the information contained in that dataset.
Claude will use the search functionality as needed to answer your questions, potentially making multiple queries to provide comprehensive responses. You can ask Claude to search for specific information, summarize content, or analyze data within your Trieve datasets.
The search functionality supports various search types and filtering options:
When working with Claude, you can specify these parameters in your natural language requests, and Claude will translate them into appropriate search queries.