graphlit/graphlit-mcp-server

npx -y graphlit-mcp-server (add to VS Code, Windsurf, Cline, Cursor, or Claude Desktop via Smithery config)
Requires env vars: GRAPHLIT_ENVIRONMENT_ID, GRAPHLIT_ORGANIZATION_ID, GRAPHLIT_JWT_SECRET

Usage

• •Configure the server in your MCP client with the three required Graphlit env vars (from the Graphlit Platform API settings dashboard)
• •Optionally add connector credentials (SLACK_BOT_TOKEN, DISCORD_BOT_TOKEN, LINEAR_API_KEY, GITHUB_PERSONAL_ACCESS_TOKEN, JIRA_EMAIL/JIRA_TOKEN, NOTION_API_KEY, etc.)
• •Ingest content (files, web pages, messages, emails, issues, text) then query contents/collections/feeds or run RAG via the Prompt LLM Conversation tool
• •On Windows, npx may need "command": "C:\\Windows\\System32\\cmd.exe /c npx"

Key features

• ✓20+ data connectors: Outlook, Google Mail, Notion, Reddit, Linear, Jira, GitHub (Issues), Google Drive, OneDrive, SharePoint, Dropbox, Box, Slack, Teams, Discord, Twitter/X, Podcasts (RSS)
• ✓Retrieval tools (Query Contents/Collections/Feeds/Conversations, Retrieve Relevant Sources/Similar Images), RAG (Prompt LLM Conversation), and Extract Structured JSON from Text
• ✓Documents (PDF, DOCX, PPTX, etc.) and HTML extracted to Markdown on ingestion; audio/video transcribed on ingestion
• ✓Built-in web crawling, web search (incl. podcast search), web mapping, and screenshot tools - no separate Firecrawl/Exa integration needed
• ✓Publishing tools (ElevenLabs audio, OpenAI image generation) and notifications (Slack, email, webhook, Twitter/X)

Best for

Building a RAG-ready knowledge base across your developer and product-management tools and querying it from an MCP client.

Caveats

• ⚠Requires an active Graphlit Platform account with an environment/organization/JWT secret (paid platform service)
• ⚠Node.js 18.x+ required
• ⚠Document connector credentials are optional; only the three Graphlit env vars are required