graphlit-mcp-server

概述

Graphlit MCP 服务器实现了 MCP 客户端与 Graphlit 服务的集成，创建了一个可搜索的、支持 RAG 的知识库，覆盖开发者和产品管理工具。它支持从多种来源获取内容，包括 Slack、Discord、网站、Google Drive、电子邮件、Jira、Linear 和 GitHub。服务器提供全面的工具，涵盖检索、RAG、提取、发布、摄取、数据连接器、Web 操作、通知和枚举功能。文档被提取为 Markdown 格式，音频和视频文件会被转录，且内置网页爬取和搜索功能，无需单独集成其他工具。

试试问 AI

装完之后，这里有 5 个你可以让 AI 做的事：

• query_contents

  Search through ingested contents in your Graphlit project

  试试:Find documents about project timelines试试:Search for information about user feedback
• query_collections

  Query across collections of content in your Graphlit project

  试试:Retrieve documents from the engineering collection试试:Find information from all product-related collections
• web_search

  Search the web for information using built-in web search

  试试:Search for latest AI trends试试:Find articles about React best practices
• web_crawling

  Crawl websites to extract their content

  试试:Crawl https://example.com/docs试试:Extract content from my company's knowledge base
• prompt_llm_conversation

  Start a conversation with an LLM using your Graphlit data as context

  试试:Summarize key findings from recent research试试:Answer questions based on my project documentation
• extract_structured_json_from_text

  Extract structured JSON data from unstructured text

  试试:Extract key information from this meeting transcript试试:Parse this customer feedback into structured format
• publish_as_audio

  Convert text to audio using ElevenLabs

  试试:Convert this summary to audio file试试:Generate voiceover from this script
• publish_as_image

  Generate images using OpenAI's image generation

  试试:Create an illustration of a concept试试:Generate a diagram showing our system architecture
• retrieve_similar_images

  Find images similar to a provided query or image

  试试:Find images similar to this product screenshot试试:Search for design patterns like this one
• visually_describe_image

  Generate a textual description of an image

  试试:Describe what's in this chart试试:Explain the contents of this screenshot
• configure_project

  Configure your Graphlit project settings

  试试:Update project settings试试:Configure ingestion options
• create_collection

  Create a new collection to organize your content

  试试:Create a collection for research papers试试:Make a new collection for customer feedback

可对比工具

安装

npx -y graphlit-mcp-server

{
  "mcpServers": {
    "graphlit": {
      "command": "npx",
      "args": ["-y", "graphlit-mcp-server"],
      "env": {
        "GRAPHLIT_ORGANIZATION_ID": "your-organization-id",
        "GRAPHLIT_ENVIRONMENT_ID": "your-environment-id",
        "GRAPHLIT_JWT_SECRET": "your-jwt-secret"
      }
    }
  }
}

FAQ

Graphlit MCP 服务器支持哪些数据源？

该服务器支持多种数据源，包括 Slack、Discord、网站、Google Drive、电子邮件、Jira、Linear、GitHub、Microsoft Teams、Twitter/X 等，通过其数据连接器实现。

是否支持不同类型的内容？

是的，它处理多种内容类型，包括 PDF、DOCX、PPTX（提取为 Markdown）、音频和视频文件（转录）、网页和结构化数据提取。

graphlit-mcp-server 对比