Launch Week is coming soon!

Each day from September 8-11, 2026 we're launching new tools, products, and integrations.

Find out more about launch week

Knowledge Base

Terminalfour AI Connector

Last Modified:
07 Sep 2026
User Level:
Power User

Connect Terminalfour to the AI tools you already use

The Terminalfour AI Connector lets AI applications interact directly with your Terminalfour system.

Using the Model Context Protocol (MCP), you can connect Terminalfour to compatible AI tools such as Claude, ChatGPT, Cursor, Kiro and Gemini CLI, then use natural language to work with your Terminalfour content and configuration.

You can use the AI Connector to explore content, create and update pages, manage site structure, work with media, manage metadata and users, and perform publishing and workflow tasks.

AI Connector is currently available through early access. Access is being provided to a limited number of institutions while we learn from real-world use cases and continue to develop the connector.

Getting started

Getting started takes just a few minutes.

1. Apply for an AI Connector account

Apply for early access by going to https://ai-connector.terminalfour.net and filling in the form

2. Log in and create a connection

Once you've been invited, you'll receive an email with instructions on how to log in.

From the AI Connector dashboard, select Create connection.

Give your connection a name and follow the guided setup.

You'll be asked to provide the details required to connect the AI Connector to your Terminalfour system. The setup will guide you through each step.

3. Choose your AI tool

Once your Terminalfour connection has been created, choose the AI application you want to use.

The AI Connector currently supports:

  • Claude
  • ChatGPT
  • Cursor
  • Kiro
  • Gemini CLI
  • Other MCP-compatible AI applications

The setup process provides the configuration instructions for your chosen application.

4. Start using Terminalfour through AI

For example:

Show me the pages in the Engineering section that were modified this week.

Or:

Create a new page for the Computer Science department using the standard department page content type.

The AI application will determine which Terminalfour capabilities it needs to complete the request and as long as the API key you provided has the rights to perform the actions, the AI tool will access Terminalfour and perform the requested actions.

How the AI Connector works

The AI Connector uses the Model Context Protocol (MCP), an open standard for connecting AI applications to external systems.

Rather than building a separate integration between Terminalfour and every AI application, Terminalfour exposes capabilities through a standard interface that compatible AI applications can use.

This means the same Terminalfour connection can potentially be used with different AI tools.

The AI application handles the conversation and decides what it needs to accomplish your request. The AI Connector provides the connection to Terminalfour and exposes the capabilities available to it.

What can I do with the AI Connector?

The AI Connector gives compatible AI applications access to Terminalfour capabilities.

The exact capabilities available to you depend on the API key you provide when creating your Connection and what rights that key has.

You can use AI to:

  • Manage content
  • Manage sections
  • Manage media
  • Manage lists
  • Manage page layouts
  • Manage users and groups
  • Manage handlebars helpers and partials
  • Publish channels, sections, and branches
  • Lookup Terminalfour documentation

Example

Find all pages in the Biology section that haven't been updated in the last 12 months.

Or:

Create a news article announcing the University's new research centre. Use the News content type and save it as a draft.

Available tools

The underlying MCP tools include capabilities such as:

Tool Name What it is used for
t4_lookup-t4-docs Used to find answers to questions about Terminalfour behavior by referencing content on the Terminalfour Docs
t4_content Create, read, update, delete, approve content. 
t4_content-types Create, read, update, and delete content types and content layouts
t4_sections Create, read, update, and delete sections. Also used for managing details on a section such as the metadata, assigned content types etc
t4_media Create, read, update, and delete media.
t4_media-categories Managing media categories
t4_media-types Create, read and update media types
t4_lists Create, read, update, and delete lists in the Terminalfour platform
t4_navigation Create, read, update, and delete navigation objects. Supports all 19 navigation object types
t4_page-layouts Create, read, update, and delete Page layouts. Can modify header and footer code
t4_system Read info about the Terminalfour system such as the version
t4_handlebars Create, read, update, and delete custom Handlebars Helpers and Partials
t4_users Create, read, update, and delete users.
t4_groups Create, read, update, and delete groups and their associated users
t4_channels Read and publish channels
t4_lookup-publishapi-docs Lookup documentation for custom handlebars code as required

Working with AI effectively

The AI Connector gives you access to Terminalfour through natural language, but the quality of the result depends on the request you give the AI.

Be specific

Instead of:

Find old pages.

Try:

Find pages in the Research section that haven't been modified in more than two years.

The second request gives the AI a much clearer set of criteria.

Start with read-only requests

When exploring what the AI Connector can do, start by asking questions that don't change anything.

For example:

Show me the Research section underneath the Main University site.

Then:

Show me the content in the Research section.

And finally:

Create a new section under the Research section using the Research News content type.

This makes it easier to understand what the AI is going to do before allowing it to make changes.

Ask the AI to explain what it is going to do

For complex operations, you can ask:

Before making any changes, tell me what you're going to change.

This can be particularly useful when making bulk changes.

For example:

Find all pages in the Student Services site without a description. Don't change anything yet. Give me the list and explain what you would change.

You can then review the result before asking the AI to proceed.

Note: the MCP tools that will perform destructive actions will always ask for explicit user confirmation.

Combining capabilities

One of the biggest advantages of connecting Terminalfour to an AI application is that you can combine multiple operations into a single request.

For example:

Find all pages in the Undergraduate Admissions site that haven't been updated in the last year. Identify which ones have missing descriptions and give me a list of the pages that should be reviewed.

A more advanced workflow could be:

Find all pages in the Undergraduate Admissions site with missing descriptions. Generate a suggested description for each page based on its content and show me the proposed changes. Don't update anything yet.

The AI can use multiple Terminalfour capabilities to complete the task rather than requiring you to perform each step manually.

Permissions and access

The AI Connector does not give an AI application unrestricted access to do everything in Terminalfour.

The actions available through a connection are governed by the permissions and capabilities associated with that connection.

This means you should treat an AI Connector connection in the same way you would treat any other integration with your Terminalfour system.

Recommended approach

Create connections specifically for the tasks you want to perform.

For example, create a connection with an API key that can only manage the parts of the platform you want the AI to manage.

Important considerations

AI can make mistakes

AI applications can misunderstand requests or make incorrect assumptions.

Always review important changes before they are applied, particularly when working with:

  • Large numbers of content items
  • Site structure
  • Metadata
  • Permissions
  • Publishing
  • User accounts

For high-impact operations, ask the AI to show you what it intends to change before asking it to make the changes.

Be careful with bulk operations

A request such as:

Update every page in the website

can have a much larger impact than expected.

For bulk operations, start with a smaller scope:

Show me the pages that would be affected in the Engineering section.

Then review the results before expanding the operation.

Don't provide sensitive information

You should not provide passwords, API keys, secrets or other credentials to an AI application as part of a prompt.

The AI Connector's connection configuration should be used for authentication and access to Terminalfour.

AI model training

Data sent through the AI Connector is not saved by Terminalfour and is not used by Terminalfour to train AI models.

However, the AI tools you connect to may have their own data retention and training policies. Check the privacy and data usage policies of the AI provider you use to understand whether conversations or data submitted through the AI Connector may be stored or used to improve their models.

Troubleshooting

My AI application can't connect

Check that:

  • Your AI Connector account is active.
  • Your Terminalfour connection has been created successfully.
  • You followed the configuration instructions exactly.
  • Your Terminalfour user has the permissions required for the requested operation.

If you've completed the setup but the connection isn't working, create a new connection and repeat the setup process.

Contact Terminalfour support if you are encountering any issues.

The AI can't perform an operation

The AI may not have access to the capability required for the request.

Try asking:

What Terminalfour capabilities are available to you?

If the required capability is listed but not working, check the permissions associated with the API key used in your connection.

The AI gives an unexpected result

AI applications interpret natural-language requests and may occasionally misunderstand what you're asking.

Try making the request more specific.

Instead of:

Update the pages.

Try:

Update the description field on the following three pages. Don't change any other fields, and show me the proposed changes before applying them.

Frequently asked questions

Do I need to be a developer?

No.

The AI Connector is designed to allow you to interact with Terminalfour using natural language. However, some use cases (particularly automation and custom integrations) may benefit from technical knowledge.

Do I need to understand MCP?

No. You don't need to understand MCP to connect or use the AI Connector.

MCP is the technology that allows your AI application to communicate with Terminalfour.

Can I use more than one AI application?

Yes. A Terminalfour connection can be used with supported MCP-compatible AI applications, allowing you to choose the AI tool that best fits your workflow.

Does the AI Connector replace the Terminalfour interface?

No.

The Terminalfour interface remains the primary way to manage your website. The AI Connector provides another way to interact with Terminalfour, particularly useful for conversational workflows, automation and tasks that involve searching or working with multiple pieces of content.

Can AI publish my website?

Publishing capabilities depend on the permissions available to your connection.

Where publishing is available, treat AI-initiated publishing in the same way you would any other publishing action: review the changes and make sure you have the appropriate permissions before proceeding.

Can AI delete content?

Capabilities depend on the permissions and tools available to your connection.

We recommend using particular care with destructive operations. By default, any destructive operations require explicit user confirmation so your AI tool should automatically confirm this with you. However, we still recommend asking the AI to show you what it intends to change before confirming the operation.

Is the AI Connector available to everyone?

The AI Connector is currently in early access and access is being provided to a limited number of institutions.

Back to top