home / mcp / knowledge base mcp server

Knowledge Base MCP Server

解决知识库搜索需要手动编排流和意图识别的问题

Installation
Add the following to your MCP client configuration file.

Configuration

View docs
{
  "mcpServers": {
    "boleyn-fastgpt-dataset-mcp": {
      "url": "http://0.0.0.0:18007/sse?parentId=YOUR_PARENT_ID",
      "headers": {
        "API_TOKEN": "your-api-token",
        "API_BASE_URL": "http://your-api-domain.com",
        "MCP_SERVER_HOST": "0.0.0.0",
        "MCP_SERVER_PORT": "18007",
        "DEFAULT_PARENT_ID": "your-default-parent-id"
      }
    }
  }
}

This MCP server provides intelligent knowledge-base search and management for FastGPT. It connects to multiple knowledge sources, enables adaptive keyword expansion, and supports cross-dataset, multi-level searches to help AI quickly locate relevant information and understand complex contexts.

How to use

You connect your FastGPT client to the knowledge-base MCP server via the SSE endpoint and start sending search queries. The server automatically expands your keywords, explores datasets, and returns organized results with detailed content when needed. Use the multi-dataset search to query across datasets and the view collection content tool to inspect full documents.

Key usage flow you should follow:

  • Expand your search terms with intelligent keyword expansion to include synonyms and related terms.
  • Discover relevant datasets by browsing the dataset tree and folders.
  • Perform precise searches within a specific dataset to retrieve exact content.
  • Run parallel searches across multiple datasets to speed up results.
  • Open full documents to view complete content and metadata when needed.

Available tools

get_dataset_tree

Fetches the knowledge-base directory tree to browse datasets and folders.

search_dataset

Performs a precise search within a specified dataset.

multi_dataset_search

Executes parallel searches across multiple datasets and aggregates results.

view_collection_content

Retrieves the full content of a document or collection item.

expand_search_keywords

Expands a query with intelligent synonyms and related terms to improve search results.

explore_folder_contents

Explores nested folders to surface deeper knowledge resources.