devflowinc/trieve
Free通过 Trieve 爬取、嵌入、分块、搜索和检索数据集信息。
FreeFree tier
Inputs: text
About devflowinc/trieve
trieve-mcp-server is a Model Context Protocol (MCP) server that provides agentic tools for AI agents to search and interact with Trieve datasets through a standardized interface. Built with TypeScript, it supports semantic, full-text, hybrid, and BM25 search across datasets, with configurable options via environment variables or command-line arguments. The server exposes datasets as MCP resources and integrates seamlessly with Claude Desktop, enabling agents to query, filter, and break down complex dataset requests. It is open-source and part of the Trieve ecosystem by Devflow Inc.
Key Features
Semantic, full-text, hybrid, and BM25 search across Trieve datasets
List and access dataset information via standardized MCP resources
Configurable via environment variables or command-line arguments
Built with TypeScript for type safety and developer experience
Seamless integration with Claude Desktop and other MCP-compatible clients
Exposes Trieve datasets as MCP resources for easy attachment and querying
Pros & Cons
Pros
- Standardized MCP interface compatible with multiple AI agent frameworks
- Multiple search types (semantic, fulltext, hybrid, BM25) for flexible retrieval
- Easy setup with environment variables or command-line arguments
- TypeScript implementation provides type safety and robust error handling
- Open source with active development and community contributions
Cons
- Requires a valid Trieve API key and organization ID for operation
- Dependent on the Trieve platform and its availability
- Limited to interacting with Trieve datasets only, not general-purpose search
- No built-in user authentication beyond the API key mechanism
Best For
Enabling AI agents to search and retrieve information from custom Trieve datasetsIntegrating Trieve search capabilities into Claude Desktop workflowsBuilding AI-powered applications that require semantic or hybrid search over structured dataAutomating dataset queries and filtering for research, support, or analytics tasks
FAQ
How do I configure the Trieve MCP server?
Configuration can be done via environment variables (TRIEVE_API_KEY, TRIEVE_ORGANIZATION_ID, and optionally TRIEVE_DATASET_ID) or by passing --api-key, --org-id, and --dataset-id as command-line arguments. Copy the .env.dist file to .env and fill in your credentials from dashboard.trieve.ai.
What search types are supported?
The server supports semantic (default), fulltext, hybrid, and BM25 search types. The searchType parameter can be set when calling the search tool.
Can I use this server with Claude Desktop?
Yes. Configure Claude Desktop's claude_desktop_config.json with the trieve-mcp-server as an MCP server, providing the required environment variables or command-line arguments. Datasets will appear as attachments that Claude can use to answer queries.
Is this tool open source?
Yes, the trieve-mcp-server is open source under the devflowinc/trieve repository on GitHub, licensed under the project's terms.