Model Context Protocol (MCP)
Introduction
The Text Model Context Protocol (MCP) server enables AI assistants to interact with Text features and data directly. This integration allows AI agents to perform real-time customer support operations and automate workflows without manual intervention.
What is MCP?
Model Context Protocol is an open standard that enables AI assistants to connect to external tools and data sources securely. Text’s MCP server is built upon our Workflows feature and provides functions that let AI agents take various actions in the Text ecosystem.
In practice, this means an AI assistant can perform tasks such as searching for tickets or listing chats inside the Text ecosystem — all within the context and permissions of a real user.
The external Text MCP currently offers a curated set of tools to ensure limited and safe access to core Text functions. As we expand our external MCP offerings, new tools may be added.
How MCP works
Text hosts a remote MCP server that follows the authenticated MCP specification. When an AI tool or application connects to this server, the interaction flows as follows:
- Connect: The AI assistant, or client application, initiates a connection to the Text MCP server at the given URL.
- Authenticate: The user is authenticated, and their identity, permissions, and scopes are verified. The MCP server enforces the same access controls as their Text account.
- Authorize and interact: Once authenticated, the AI assistant can invoke the available tools (API-like actions) to retrieve or manipulate Text data within the user’s permitted scope.
- Live connection: The MCP connection remains open to stream results and receive real-time updates as needed. This allows AI assistants to maintain context and respond promptly.
Connecting to Text MCP
We support out-of-the-box, browser-based MCP connection to Claude and ChatGPT through the https://mcp.text.com/
endpoint. When logged in to your Text account, these AI assistants will get instant access to your data within available user scopes.
Claude
You can connect your Claude to the Text MCP in Claude settings by pasting the MCP URL.
ChatGPT
To connect ChatGPT, add a new connector in ChatGPT settings and paste the MCP URL. Note that, at the moment, custom connectors for ChatGPT are only available in the Pro, Team, Enterprise and Edu plans and support only fetch and search tools. To use MCP features, you must activate deep research.
Using MCP-remote
The Text MCP external server supports public authentication for all compatible MCP clients.
External clients connect via mcp-remote
to the public endpoint:
MCP URL |
---|
https://mcp.text.com |
To start mcp-remote
, you must specify a local port. The MCP client is then configured to connect to the local address using the chosen port.
You can connect to Text MCP through the following ports:
6274
53531
59123
60222
Configuration example
Below is a generic example of an mcp-remote
OAuth connection. Setup details may vary between versions and providers, so always refer to your specific LLM provider's official documentation for the most up-to-date configuration instructions.
Browser-based OAuth
{
"mcpServers": {
"text": {
"command": "npx",
"args": [
"mcp-remote",
"https://mcp.text.com",
"<port>"
]
}
}
}
User scopes
The Text MCP server enforces the authenticated user’s existing Text account permissions, restricting all actions and data access to their assigned roles, teams, and authorized resources. This ensures AI assistants connected through MCP operate under the same security model as the authenticated user.
Available MCP tools
The external Text MCP server exposes a predefined set of tools that an AI assistant can use. These tools correspond to common support operations inside Text and often combine multiple logics from multiple endpoints to fetch and act on data.
The below tools represent the current external MCP capabilities. They map directly to underlying Text APIs and mirror their behavior and parameters.
Common
Method | Description |
---|---|
search | Search for data in the Text ecosystem. |
fetch | Retrieve data for specific resources. |
Chats
Method | Description |
---|---|
get-chat-transcript | Retrieve the full transcript of a chat. |
list-archived-chats | List archived chats with filtering and pagination. |
list-chat-tags | List tags from a chat. |
Tickets
Method | Description |
---|---|
find-ticket | Find an existing ticket by its ID. |
list-tickets | List tickets with pagination and optional filtering. |
list-ticket-tags | List tags from a ticket. |
Contact us
If you still can't find the answer you're looking for, don't hesitate to contact us at platform@text.com or write on our Discord server.