Graphlit MCP Server

How to use

Once the MCP server is running, you connect your MCP client (such as Cursor, Windsurf, Goose, or Cline) to the server endpoint. Ingest data from your sources (Slack, Discord, websites, email, Jira, GitHub, etc.) and organize it into Collections and Feeds. Use the server’s tools to search, retrieve, and refine results, including web crawling, extraction, and RAG-style conversations. The goal is to build a searchable, knowledge-rich base that your MCP clients can query for relevant information, across documents, pages, and conversations.

Typical usage patterns include: - Ingest data from connected sources into a Graphlit project. - Create and configure Collections and Feeds to organize ingested content. - Run web crawls and web searches to supplement your knowledge base. - Extract structured JSON from text to feed downstream workflows. - Publish outputs as audio or images when needed. - Use notifications to keep your team informed about new content or completed tasks.

How to install

Prerequisites you need before starting: - Node.js version 18.x or higher. - An active Graphlit Platform account with API access.

Option A: Quick install via one-click in your editor (VS Code) — choose one of the provided install buttons and follow the prompts. The setup will configure and run the MCP server for you.

Option B: Manual install by configuring your MCP client settings and running the server with NPX.

Option C: Install with Windsurf, Cline, Cursor, or Smithery as shown in the examples below.

Configuration and usage notes

Configure the MCP server to authenticate with the Graphlit platform using environment variables. The following values are required for a basic setup: - GRAPHLIT_ORGANIZATION_ID: Your Graphlit organization ID. - GRAPHLIT_ENVIRONMENT_ID: Your Graphlit environment ID. - GRAPHLIT_JWT_SECRET: Your JWT secret used to sign tokens.

Optional credentials for data connectors can be added as needed (Slack, Google Email, Notion, Jira, GitHub, etc.). These are not required for a minimal setup but enable broader ingestion capabilities.

Examples of typical MCP server configurations

{
  "mcpServers": {
    "graphlit-mcp-server": {
      "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"
      }
    }
  }
}

Security and maintenance notes

Store sensitive values (organization ID, environment ID, JWT secret) securely and avoid committing them to source control. Rotate JWT secrets periodically and follow your organization’s security practices for token management.

Query Contents

Retrieve content items that match a query across all ingested sources.

Query Collections

Fetch contents within a specific collection for targeted results.

Query Feeds

Search within data feeds to locate the most relevant sources.

Query Conversations

Interact with stored conversations and retrieve context.

Retrieve Similar Images

Find visually similar images from ingested media.

Web Crawling

Crawl and map web pages to add fresh content to your knowledge base.

Web Search

Perform web searches, including podcast search, to expand coverage.

Extraction

Extract structured JSON from text to enable downstream processing.

Publish as Audio

Publish generated content as audio using ElevenLabs or similar services.

Publish as Image

Publish generated content as images using image generation services.

Ingestion — Files

Ingest files from local or connected storage sources.

Ingestion — Web Pages

Ingest and convert web pages to Markdown for storage.

Notifications

Notify via Slack, Email, Webhook, or Twitter/X when events occur.

Configure Project

Set up and adjust project-level settings and workflows.

Create Collection

Create new collections to organize ingested content.

Add Contents to Collection

Add new content to existing collections.

Remove Contents from Collection

Remove items from a collection as needed.