This guide walks through connecting ChatGPT to the Hudu MCP (Model Context Protocol) Server.
Once connected, ChatGPT can interact directly with your Hudu instance to:
- Read documentation
- Create articles
- Update articles
- Review activity logs
- Assist with documentation workflows
Requirements
Before starting:
- MCP access must already be enabled in Hudu
- Your Hudu instance should be accessible over HTTPS/SSL
- You must have access to ChatGPT Apps/Developer Mode
Your MCP endpoint will typically look like:
https://yourhuduinstance.com/mcpEnable MCP in Hudu First
Inside Hudu:
- Navigate to:
Admin → External Apps → Model Context Protocol (MCP)- Enable MCP access if it is not already enabled.
- Confirm your MCP endpoint URL.
Setup in ChatGPT
Step 1. Open ChatGPT Settings
Inside ChatGPT:
- Click into your profile
- Click Settings
- Open Apps
- Enable Developer Mode
Depending on your environment, you may see warnings about unverified apps. This is expected for privately hosted MCP servers.
Step 2. Create a New App
- Click Create App
- Enter:
- App Name
- Description
- MCP Server URL
Example:
https://yourhuduinstance.com/mcpStep 3. Configure OAuth
- Select OAuth authentication
- Continue through the warning prompts
- Click Create
ChatGPT will now attempt to connect to your Hudu MCP server.
Step 4. Authenticate with Hudu
After the app is created:
- A Hudu OAuth login window will appear
- Sign in to Hudu
- Approve the requested permissions
Once approved, ChatGPT will automatically discover the available Hudu MCP tools.
OAuth Authorization
When connecting to the Hudu MCP server for the first time, Hudu will display an OAuth authorization screen asking the user to approve the connection.
The authorization request includes the required permissions (scopes), including:
- Read access
- Write access
Users must approve the requested scopes before the MCP connection can be established successfully.
Verifying the Connection
After authentication:
- ChatGPT should display available Hudu tools
- You can begin interacting with Hudu using natural language prompts
Example prompts:
- “Create a draft SOP for workstation onboarding”
- “Find documentation related to backup failures”
- “Summarize recent documentation changes”
- “Update this article with the latest firewall procedure”