> ## Documentation Index
> Fetch the complete documentation index at: https://docs.tranthor.com/llms.txt
> Use this file to discover all available pages before exploring further.

# Connect Claude or ChatGPT

> Work in your Tranthor workspace from Claude or ChatGPT: what it can read, what it can change, and how sending stays in your hands.

An Assistant connection lets you use your workspace from Claude or ChatGPT in plain language. Ask how a broadcast performed, have a draft written, or send a small broadcast, without opening the app.

The assistant acts as **you**. It can do only what your role in the workspace allows, and what it builds is recorded as your work, done through the assistant.

## Connect

**1. Add Tranthor in Claude**

In Claude, open **Settings**, then **Connectors**, and add Tranthor from the directory. To add it by hand, choose **Add custom connector** and enter the server address `https://api.tranthor.com/mcp`.

**2. Sign in to Tranthor**

Sign in with your Tranthor account when asked.

**3. Choose a workspace and access**

Pick the one workspace the assistant will work in, and choose what it may do (see below). Select **Allow**.

Each connection works in the one workspace you choose. To use another workspace, or to change the access you gave, connect again.

## What the assistant can do

| Access | What it covers |
| - | - |
| **Read** | Segments, journeys, broadcasts and message templates. The last 30 days of revenue, orders and messages (the numbers on Home). Analytics on delivery, engagement, orders and conversions, as totals and breakdowns. What is waiting for you in **Needs you**. Always on. |
| **Make changes** | Save draft segments, journeys, broadcasts and message templates. Hand a brief to Tranthor to build from. Pause and resume a journey that has already run. Ask the Supervisor. Offered when your role allows it. |
| **Send** | Send a draft broadcast or start a journey that has never run, after two confirmations. Optional, and off unless you tick it. |

A draft does nothing until someone sends, starts or publishes it. The assistant cannot publish, approve or delete anything.

## Roles

| Your role | What the assistant can do |
| - | - |
| Viewer | Read only. |
| Author | Read, and save drafts, hand a brief to Tranthor, and ask the Supervisor. |
| Workspace manager, Admin | Everything an Author can do, plus pause and resume journeys, and send. |

## Sending takes two confirmations

Before anything reaches customers, the assistant shows you Tranthor's preview: how many customers it reaches, the sender, and any risks. You approve the preview, then type the number of customers yourself (for a journey that starts on an event, its name). Only then does the send go out. The assistant never fills in the number on its own.

Some sends are never made from an assistant. A send that reaches your whole audience, more customers than you have approved before, or a new channel or sender is **Critical**. A person approves it inside Tranthor.

Resuming a paused journey sends again to customers who are already in it, so the assistant tells you so and asks first.

## Keeping control

* **See what it did.** Settings, **AI Agents** lists each connection and its activity log.
* **Revoke any time.** Revoking it in Settings, **AI Agents** ends the access straight away. It also ends if you leave the workspace or your role drops below what you allowed.
* **Checked on every call.** Each request re-checks the connection, your membership and your role.

## Privacy

An Assistant connection returns counts, results, drafts and links into the app. It does not return individual customers' contact details. Tranthor keeps a log of what each connection did, and the connection itself, until you revoke it. See the [privacy policy](https://tranthor.com/privacy).

## Troubleshooting

* **Your assistant asks you to sign in again.** Connect again from your assistant's settings, and keep the workspace and access you want.
* **You do not see the tools to save drafts or send.** Your role does not allow them, or you did not allow them when you connected. Connect again and choose the access you need.
* **A send is refused as Critical.** Open the link the assistant gives you and approve it in Tranthor.
* **You need help.** Write to [support@tranthor.com](mailto:support@tranthor.com).

AI outputs may contain mistakes. Check important information before you act on it.

## Related

* [MCP server reference](https://docs.tranthor.com/guide/mcp-server.md): the server address, sign-in, access and tools, for developers.
* [AI Agents](https://docs.tranthor.com/guide/ai-agents.md): where Tranthor's own agents live and how you stay in control.
* [Contact controls](https://docs.tranthor.com/guide/contact-controls.md): the limits every agent send must respect.


This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.