Connect Trainn to your AI assistant to find answers in your training content and create new training through chat.
Setup has four parts: enable MCP in Trainn, add the connection in your assistant, approve access, and check the connection.
Before you start
You need:
A Trainn account with access to the specific workspace you want to connect.
A Trainn role that allows the Library or Academy actions you want to perform.
Permission to add custom connections in your AI workspace. Your organization’s administrator may need to enable this.
Trainn MCP server URL:
https://mcp.trainn.co/mcpThe above URL should be entered inside your AI Assistants MCP connector settings to setup a connection with Trainn.
1. Enable Trainn MCP
A Trainn workspace administrator must complete this step before anyone can connect an assistant.
In Trainn, open Settings → Integrations → Trainn MCP.
Select Enable.
Copy the MCP server URL.
Enabling MCP makes the connection available for your workspace.
2. Add Trainn to your AI assistant
Use the same Trainn server URL for each assistant.
Claude
Open Claude and go to Customize → Connectors.
Use the add button to add a custom connector.
Name it Trainn and enter
https://mcp.trainn.co/mcp.Follow the connector setup prompts and review the detected authentication settings.
Complete Trainn sign-in and permission approval as described in step 3 below.
Start a chat and enable Trainn in the conversation’s connectors.
If your organization manages Claude connectors, an authorized owner must first add Trainn under the organization’s connector settings. Members can then find Trainn in their connectors and sign in individually.
See Claude’s remote connector setup guide for the current screens and organization controls.
ChatGPT
Use ChatGPT on the web. Your workspace’s permissions and security settings must allow custom MCP connections.
Open Plugins, use the plus button, and choose Add custom MCP server.
Name it Trainn and enter
https://mcp.trainn.co/mcpas the server URL.Configure OAuth authentication, review the displayed warning, and create the plugin.
Install the resulting plugin from your personal or workspace plugins.
Complete Trainn sign-in and permission approval when prompted.
After connecting, start a new chat and make sure Trainn is enabled in the tools panel.
See OpenAI’s custom MCP setup guide for the current connection flow.
Codex — if you already have the Trainn plugin installed
Open the installed Trainn plugin and complete its connection flow.
Sign in to Trainn and approve the intended workspace and permissions.
Select Trainn in the chat and run the connection check below.
Check the Codex connection separately, even if Trainn already works in ChatGPT. When an in-chat upload panel is unavailable, use the secure Trainn upload-page link your assistant provides.
Notion
MCP connections require a Notion Business or Enterprise plan. You can set them up on web or desktop.
Connect your personal Notion Agent
Open Settings → Connections → Discover.
Select Add Custom MCP. If this option is unavailable, ask your workspace admin to allow and configure the custom connection.
Enter the Trainn server URL and name the connection Trainn.
Complete Trainn sign-in and approve the workspace and permissions.
Find the connection in your chat’s MCP sources and run the connection check below.
Your personal connection uses your Trainn account. Other teammates need their own connections. See Notion’s personal Agent connection guide.
Steps to Connect a Notion Custom Agent
Ask a Notion administrator to enable custom MCP servers and allow the Trainn connection.
Open the agent’s settings and its Tools & Access section.
Add a connection and choose the custom MCP option.
Enter the Trainn server URL, name the connection Trainn, and complete sign-in.
Make sure the Trainn tools the agent needs are enabled.
Custom Agents have their own connections. Each uses the credentials and permissions of the person who authenticated it. People allowed to interact with the agent may invoke its connected tools, so review the agent’s sharing settings.
See Notion’s Custom Agent connection guide for the admin and agent setup steps.
Cursor
To connect with Cursor Local instance
Open Settings → General → Open Customize
Click MCPs from the navigation bar on top.
Click New MCP Server
Add the following Trainn entry to
.cursor/mcp.jsonin your project root. If the file already contains other servers, keep those entries.{ "mcpServers": { "Trainn": { "url": "https://mcp.trainn.co/mcp" } } }Save the file and restart Cursor.
Once added, click Trainn, and choose Authenticate
Complete the Trainn sign-in and approval flow in your browser.
Open an Agent chat and run the connection check below.
See Cursor’s MCP setup guide for configuration help.
If your Cursor version supports only local command-based MCP servers, use the bridge configuration below. It requires Node.js 22.16 or later.
{
"mcpServers": {
"Trainn": {
"command": "npx",
"args": ["-y", "mcp-remote@latest", "<https://mcp.trainn.co/mcp>"]
}
}
}
Restart Cursor after configuring it and complete the Trainn sign-in flow.
3. Sign in and approve access
During connection, Trainn opens in your browser.
Sign in with your Trainn account.
Select the workspace you want the assistant to use.
Review the requested permissions.
Select Allow.
Your assistant works in the selected workspace. Your Trainn role still applies to every action.
Choose the access you need
These are the labels on the Trainn consent screen. They come from the scope registry the authorization API returns. Your existing Trainn role still applies to every action.
Permission | What it allows |
|---|---|
Browse Library content | Search and read Library videos, guides, articles, and documents, including drafts. |
Add Library content | Upload or import videos and PDFs, create or edit Library articles, update content details, publish Library content, and delete content after confirmation. |
View courses | View courses, drafts, chapters, lessons, quizzes, settings, and preview links. This is the default permission. |
Create and manage courses | Create, edit, delete, publish, and unpublish courses. Manage chapters, lessons, quizzes, settings, and attached content. |
Add Library content includes more than uploads. Review the full access described above before approving it.
Publishing, unpublishing, and deleting courses still require an explicit request.
If an action needs a permission you did not approve, reconnect Trainn and approve that permission. Your Trainn role must also allow the action.
4. Test your connection
Step 1: Start a chat with Trainn enabled and ask:
"Who am I connected as in Trainn? Show my workspace and permissions."
A successful response includes your Trainn account, selected workspace, and granted permissions.
If you have Academy access, you can also ask:
"List my Academy courses by most recently updated."
Step 2: Try one of these requests:
Search your Library
"Search my Trainn Library for (mention any topic(s))"
Build a course — requires Academy and the relevant permissions
"Build a course on the following topic (mention any topic based on the content already present inside Trainn)"
For more workflows and example prompts, see Create training and find answers with Trainn MCP.
Troubleshooting
1. Trainn says the connection is unavailable
Ask a Trainn workspace administrator to confirm MCP is enabled. Trainn blocks the connection while MCP is disabled.
2. Sign-in fails or you see “Unauthorized”
Sign in to Trainn in the same browser, then try connecting again.
Disconnect and reconnect Trainn in your AI assistant.
Check that you selected the intended workspace.
If Academy permissions are unavailable, confirm Academy is enabled and your Trainn role permits the action.
3. An action or tool is missing
Your assistant shows only the tools allowed by the permissions you approved. Reconnect and approve the missing permission if your Trainn role allows it.
Academy tools are unavailable when Academy is disabled in the workspace.
4. Trainn does not appear in the assistant
Restart the assistant, start a new chat, and check its tools or connections.
For Notion, check that your admin allows custom MCP servers. For a Custom Agent, also check that Trainn is connected to that specific agent.
5. Your organization blocks custom connections
Ask the administrator of your AI workspace to review and allow the connection. Enabling MCP in Trainn does not override restrictions in your AI assistant’s workspace.
6. What happens if MCP is disabled?
Disabling Trainn MCP disconnects every assistant connected to that Trainn workspace. After an administrator enables it again, members must reconnect their assistants and approve access again.
Contact Support
In case you find the above troubleshooting guidelines not working, contact support@trainn.co.
