Model Context Protocol (MCP) Server for Graphlit Platform

The Graphlit MCP Server facilitates integration between MCP clients (like Cursor, Windsurf, or Cline) and the Graphlit service, enabling knowledge retrieval within these clients. It supports ingestion from various sources such as Slack, Discord, websites, Google Drive, email, Jira, Linear, and GitHub, converting documents and HTML to Markdown, and transcribing audio/video files.

Key features include retrieval (querying content/collections, finding relevant sources/similar images), extraction (structured JSON from text), and ingestion (files, web pages, messages, etc.). It offers data connectors for platforms like Microsoft Outlook, Google Mail, Notion, Reddit, Linear, Jira, GitHub Issues, Google Drive, Slack, and Discord. Web crawling and search are integrated, eliminating the need for separate tools.

Installation instructions are provided for Windsurf, Cline, Cursor, Smithery, and manual setup, all using npx -y graphlit-mcp-server. Configuration requires setting environment variables (GRAPHLIT_ORGANIZATION_ID, GRAPHLIT_ENVIRONMENT_ID, GRAPHLIT_JWT_SECRET).