Connecting Guru's MCP Server

The Guru MCP (Model Context Protocol) Server lets you connect external AI tools—like Claude, Cursor, or any MCP-compatible client—directly to your Guru workspace. This allows those tools to search, ask questions, retrieve content, and even update Cards using Guru’s Knowledge Agents. This guide walks you through setting up and using the MCP Server, including authentication, integration steps, troubleshooting, and common use cases.

👥

Access Required

This feature is currently in early access but is available using the below instructions.

💬 We want to hear from you! Share any feedback with our team about your experience using our MCP Server here.

Understanding the MCP Server

The Guru MCP Server is a protocol-based interface that allows AI tools to interact with Guru's knowledge base using Knowledge Agents. Our Guru MCP Server can:

  • List Knowledge Agents – View all available Knowledge Agents in your workspace.
  • Ask Questions – Query your Knowledge Agents and receive AI-powered answers.
  • Search Content – Perform searches across your Guru knowledge base.
  • Create a Card Draft – Generate draft Cards from within AI tools.
  • Update a Card – Push updates to existing Cards (bypassing draft creation).

By establishing this connection, users leverage all content connected to Guru— not just Guru Cards, but also your connected Sources (like Salesforce, Slack, Google Drive, Confluence, and SharePoint.) This integration enhances productivity by enabling AI tools to surface accurate, real-time knowledge from across the organization.

With support for both OAuth (currently approved for Claude, Cursor, and ChatGPT) and API token-based authentication, the MCP Server provides flexible setup options to suit various tools and automation platforms. Additionally, you can create and update Guru Cards directly from your connected clients, streamlining content management.

While currently in early access, the MCP Server focuses on core use cases such as knowledge retrieval and basic content creation to validate key workflows. Most teams can get up and running in minutes, making it a powerful, lightweight solution for extending Guru’s value into external AI workflows.

The below diagram illustrates the flow of communication between Sources, the MCP Server, and MCP Clients:


Exploring MCP Server use cases across teams

For Developers (e.g., Cursor)

  • Access and reference technical documentation while coding.
  • Query internal APIs and architecture standards.
  • Pull implementation examples from Guru Cards.

For Content Creators (e.g., Claude and ChatGPT)

  • Reference company style guides and brand standards.
  • Look up product information and specs.
  • Maintain consistency using trusted internal sources.

For Automation Workflows

  • Build automations that pull from verified knowledge.
  • Power chatbots with access to your internal documentation.
  • Develop tools that use company expertise intelligently.

Setting up your MCP Server connection

MCP Server URL: https://mcp.api.getguru.com/mcp

Step 1: Choose your authentication method

Option A: OAuth (Recommended for Claude & Cursor & ChatGPT)

  • Pre-approved for Claude and Cursor and ChatGPT
  • Other applications require whitelisting.
❗️

Important

Your application must be whitelisted. To approve your OAuth client: Contact Guru Support.

Option B: API Token

  • Works with any MCP-compatible client.
  • Ideal for automation platforms or iPaaS integrations.

Step 2: Connect your application

For Claude users

  1. Open Claude Desktop settings.
  2. Navigate to Connectors.
  3. Add Guru.
    1. NOTE: If Guru is not listed as an available connector, reach out to your company's Claude Admin to add it.
  4. Authenticate using OAuth.
  5. Enable each tool and select your permissions for each Tool.
  1. Start a new chat and hit the search and tools button(two lines), and ensure your Guru connector is toggled on

For ChatGPT users:

  1. Access ChatGPT Connector Setttings
  2. Select Guru
    1. NOTE: If Guru is not listed as an available connector, reach out to your company's ChatGPT Admin to add it.
  1. Select Connect and approve the use of the Guru connector
  1. Start a new chat, hit the + and select Use Connectors, then click the Sources drop down and ensure your Guru connector is toggled on

🚧

NOTE: The Guru connector to ChatGPT currently ONLY supports searching for content. Additional functionality (edit and draft Guru content) is pending approval but will hopefully be available soon.

For Cursor users

  1. Access Cursor settings.
  2. Go to MCP connections.
  3. Add Guru as a provider.
  4. Complete OAuth authentication.
  5. Configure Knowledge Agent access.

For other applications

  1. Generate an API token from Guru Admin settings.
  2. Configure your MCP client with the Guru MCP endpoint.
  3. Use the API token for authentication.

Step 3: Start using Guru in your workflow

Once integrated, your connected tool can:

  • Ask questions about your internal documentation.
  • Search for specific topics across your knowledge base.
  • Integrate Guru’s AI capabilities into your development or writing workflows.


Troubleshooting common issues

What does this OAuth error mean?redirect_uri parameter is invalid**

Your application must be whitelisted. To approve your OAuth client: Contact Guru Support.

Connection issues

  • Verify your API token is active and has correct permissions.
  • Ensure your client supports the MCP standard.
  • Check your network connection to Guru’s servers.

Getting support and providing feedback

During the early access phase:

  • OAuth Whitelisting – Contact support to approve your application.
  • Technical Issues – Send logs and error messages to Guru support.
  • Feature Requests – Share your use cases and feedback to help improve the MCP Server.