Connecting the TextUnited MCP Server to Your Favorite AI Tools

TextUnited now exposes an MCP (Model Context Protocol) server, which means you can bring TextUnited's terminology and asset management directly into the AI tools you already use for coding and content work — think Codex, Cursor, Cloud Code, or any other MCP-compatible application. Once connected, you can simply ask your assistant to look up or manage terminology without leaving your workflow.

Here's how to get set up.

1. Generate a Personal Access Token

Before you configure anything on the application side, you need a token from the TextUnited portal to handle authentication.

Head to [Account Settings](https://app.textunited.com/account-settings?tab=settings) and click **Create token**.

When creating the token, you'll be asked to define its scope — this controls what the token is allowed to do. For terminology access, choose the **assets** scope:

- **assets read** — allows viewing terminology only
- **assets write** — allows creating and modifying terminology

Copy and save the token right after it's generated. For security, TextUnited only stores a signature/hash of the token, not the plain text version — so once you close that screen, the token cannot be retrieved again.

2. Configure the Connection

The exact setup screen differs from app to app, but the underlying connection details are always the same:

- **Server URL:** `https://mcp.textunited.com/mcp`
- **Transport Type:** HTTP (this tells the application the server is remote, not running locally)
- **Authorization:** Bearer token, passed as an HTTP header:
`Authorization: Bearer <your_token>`

3. Application-Specific Setup

**UI-based tools (e.g., Codex/ChatGPT app):** Go to Settings, find the "Plugins and MCP servers" section, add a new server, and enter the URL along with the authorization header.

**CLI-based tools (e.g., Cloud Code):** Run the appropriate command, specifying the transport (`http`), a name for the connection, the server address, and your bearer token.

Check your application's own documentation for the exact field names or command syntax, since these vary by tool.

4. Verify the Connection

After saving your configuration, restart the application. It can take anywhere from one to seven minutes for the app to recognize the TextUnited MCP server as active.

To confirm it's working, simply ask your AI assistant whether it can see the TextUnited MCP server — if the connection succeeded, it will confirm access and you're ready to go.

Once connected, your AI assistant can tap into TextUnited's terminology directly inside your everyday tools — no more switching tabs to check a term or update your glossary.