> For the complete documentation index, see [llms.txt](https://docs.dapta.ai/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://docs.dapta.ai/dcal/developers/connect-with-sign-in.md).

# Connect Claude or ChatGPT by signing in

Connect Claude, ChatGPT, Claude Code, Codex, Cursor, OpenClaw, Hermes or another app that supports MCP sign-in to dCal by signing in with your Dapta account, with no API key to paste. Who can connect,

Claude, ChatGPT, Claude Code, Codex, Cursor, OpenClaw, Hermes and other apps that support MCP sign-in connect to dCal's MCP server, at `https://mcp.dcal.ai/mcp`, by signing in with your Dapta account: the same account you use for dCal. There is no key to create, copy or paste. This page shows who can connect, the steps for each app, and how to disconnect.

> **💡 Tip:** New to MCP? [Connect an AI assistant (MCP)](https://docs.dapta.ai/dcal/developers/mcp-server.md) explains what it is, and the other way to connect, with an API key.

***

## Sign in or use an API key

|                | Sign in                                                                                               | API key                                                                                                    |
| -------------- | ----------------------------------------------------------------------------------------------------- | ---------------------------------------------------------------------------------------------------------- |
| Made for       | Claude, ChatGPT, Claude Code, Codex, Cursor, OpenClaw, Hermes and other apps that support MCP sign-in | Agents and scripts that run with no one to sign in, and any agent that sends a header                      |
| What you paste | Nothing: you sign in with your Dapta account                                                          | A key, kept in an environment variable                                                                     |
| What it can do | Everything your owner or admin access allows: every tool, on every event type                         | Only the access you tick, for example read only or a single event type, with an expiry date if you set one |
| How it ends    | You remove the connector in the app, or an owner removes your admin role                              | An owner or admin revokes the key, or it expires                                                           |

Signing in is the quickest way. Use an [API key](https://docs.dapta.ai/dcal/developers/api-keys.md) when the assistant should have narrower access.

***

## Before you start

* You are an **owner** or **admin** of the dCal workspace. Members cannot connect an app by signing in.
* You have opened dCal on the web, at [app.dcal.ai](https://app.dcal.ai), in that workspace within the last 30 days.
* You sign in with the same Dapta account you use for dCal.

***

## Find the steps in dCal

**Step 1:** In the left menu, click **Connect your agent**, or go straight to [app.dcal.ai/admin/developers/connect](https://app.dcal.ai/admin/developers/connect). The banner and the **Connect your agent** card on **Home** open it too, and so do older links to [app.dcal.ai/admin/developers](https://app.dcal.ai/admin/developers) and [app.dcal.ai/admin/settings/developer](https://app.dcal.ai/admin/settings/developer).

The page opens with **Connect Claude, ChatGPT, Codex or Cursor to dCal** and a card, **Connect with your account**, marked **Recommended**. Its three points sum up how a sign-in works: **No API key.**, **One connector, all your workspaces.** and **Your access, nothing more.** A **Tools** button at the top right lists what a connected agent can do.

<figure><img src="https://3835013762-files.gitbook.io/~/files/v0/b/gitbook-x-prod.appspot.com/o/spaces%2FCy5rSNtQmtqYCGzJlNEB%2Fuploads%2Fgit-blob-009bc3880f072ad95498e46102418257d2e20127%2Fdcal-mcp-server-01-connect-your-agent.png?alt=media" alt="The top of the Connect your agent page: the MCP eyebrow, the heading Connect Claude, ChatGPT, Codex or Cursor to dCal, its intro and a Tools button at the top right, then the Connect with your account card with its Recommended badge, the three points No API key., One connector, all your workspaces. and Your access, nothing more., the line Pick your app and follow the steps. and the tabs Claude, ChatGPT, Claude Code, Codex, Cursor, OpenClaw and Hermes"><figcaption><p>One page, one tab per app. None of the tabs asks for a key.</p></figcaption></figure>

***

**Step 2:** Under **Pick your app and follow the steps.**, choose your app's tab: **Claude**, **ChatGPT**, **Claude Code**, **Codex**, **Cursor**, **OpenClaw** or **Hermes**. Each tab opens with a short animation of that app connecting and booking a meeting, then numbered steps with what to paste: the MCP address under **MCP URL** (**Copy URL**), or the command or configuration your app needs, each with a copy button.

The last step of every tab, **Ask**, holds **Try it**: a first request to paste into your app, with **Copy prompt**. Under the tabs, **One connector for every workspace.** reminds you that one connection works in every workspace where you are an owner or admin. At the bottom of the page, **What you can do with the MCP** sums up the tools, with a link, **See every tool**, to the **Tools** page.

On a phone, the page asks you to open it on your computer for the steps that need one, such as copying a command.

***

## Claude (claude.ai and Claude Desktop)

The **Claude** tab has buttons that open Claude on the web (**Open Claude**) or in Claude Desktop (**Open Claude Desktop**).

**Step 3:** In claude.ai, Claude Desktop or Cowork, click **+** in the chat box, then **Connectors**, **Add connector** and **Add custom connector**.

**Step 4:** Use `dCal` as the name and paste the MCP address, `https://mcp.dcal.ai/mcp`. Leave the advanced settings empty and click **Add**.

**Step 5:** Your browser opens the sign-in page. Sign in with your Dapta account. Back in Claude, dCal shows as **Connected**. Open a new chat and ask, for example, what meetings you have this week in dCal.

***

## ChatGPT

The **ChatGPT** tab has an **Open ChatGPT** button.

**Step 3:** In the ChatGPT desktop app, switch to **Work** mode. Open **Plugins**, click **Add** and choose **Create MCP App**.

**Step 4:** Fill in the form. Name: `dCal`. Server URL: `https://mcp.dcal.ai/mcp`. Leave Authentication on **OAuth**, tick **I understand and want to continue** and create it.

**Step 5:** Click **Continue**. Your browser opens: sign in if it asks, then click **Open ChatGPT** to come back.

**Step 6:** In **Plugins**, open the **Personal** tab, choose **dCal** and click **Install plugin**. The browser opens once more to link your account: click **Open ChatGPT** again. That second sign-in is expected.

**Step 7:** Click **Try now**, type `@dCal` and ask.

***

## Claude Code

This is Claude Code's own sign-in flow: [Claude Code's MCP documentation](https://docs.claude.com/en/docs/claude-code/mcp) has the details.

**Step 3:** Run this once in a terminal:

```bash
claude mcp add --transport http --scope user dcal https://mcp.dcal.ai/mcp
```

`--scope user` adds dCal for all your projects, and `dcal` is the name Claude Code gives the server. If you added dCal earlier under another name, such as `dapta-calendars`, that one keeps working. If you already added dCal with a key under the name `dcal`, pick another name for one of them.

**Step 4:** Open Claude Code, run `/mcp`, choose **dcal** and then **Authenticate**. Sign in in the browser with your Dapta account and come back to the terminal.

Already connected dCal in Claude? Claude Code signed in with the same Claude account uses that connector too: run `/mcp` to check.

***

## Codex

**Step 3:** In a terminal, run:

```bash
codex mcp add dcal --url https://mcp.dcal.ai/mcp
```

**Step 4:** Run this command. Your browser opens: sign in and come back to the terminal.

```bash
codex mcp login dcal
```

**Step 5:** Open Codex and ask.

Using the Codex or ChatGPT desktop app? Follow the steps for [ChatGPT](#chatgpt).

***

## Cursor

**Step 3:** With Cursor installed, click **Add to Cursor** on the tab. Cursor opens and asks you to confirm; the link carries only the server's name and the MCP address. Or, in Cursor, open **Customize**, **MCPs**, **New MCP Server**, choose **User** and paste this into `mcp.json`:

```json
{
  "mcpServers": {
    "dcal": {
      "url": "https://mcp.dcal.ai/mcp"
    }
  }
}
```

**Step 4:** **dcal** shows up with **Needs authentication**. Click **Authenticate**, sign in in the browser, click **Allow access** and then **Open Cursor**.

**Step 5:** Open a **New Chat** and ask. Cursor asks before it runs a tool: click **Run**.

***

## OpenClaw

**Step 3:** In a terminal, save dCal as a remote MCP server that signs in with your account:

```bash
openclaw mcp set dcal '{"url":"https://mcp.dcal.ai/mcp","transport":"streamable-http","auth":"oauth"}'
```

**Step 4:** Run this command. Your browser opens: sign in and come back to the terminal. `openclaw mcp status --verbose` then shows dcal as authorized.

```bash
openclaw mcp login dcal
```

**Step 5:** Ask your OpenClaw agent.

***

## Hermes

**Step 3:** Open `~/.hermes/config.yaml` (or run `hermes config edit`) and add dCal under `mcp_servers`:

```yaml
mcp_servers:
  dcal:
    url: "https://mcp.dcal.ai/mcp"
    auth: oauth
```

**Step 4:** Run this command. Your browser opens: sign in and come back to the terminal. Then start a new session, or run `/reload-mcp` in the one you have open.

```bash
hermes mcp login dcal
```

**Step 5:** Ask your Hermes agent.

***

## Other apps that support MCP sign-in

Any other app that can sign in to an MCP server needs only the address. Add `https://mcp.dcal.ai/mcp` as a connector or MCP server with no key, choose sign-in (often called OAuth) if the app asks how to authenticate, and sign in with your Dapta account when the app asks. Follow the app's own instructions for where to add it.

***

## Check the connection

On **Connect your agent**, under **Try it**, click **Copy prompt** and paste the prompt into your app:

> What meetings do I have this week in dCal?

The prompt only reads, so it never books anything. To see which workspace and which access the connection has, ask the assistant which workspace it is connected to: it calls the `whoami` tool and names the workspace, its owner and what it may do. A sign-in shows up there with a name that starts with `login:` instead of a key prefix, and with all five kinds of access.

***

## If you belong to several workspaces

A sign-in reaches every dCal workspace where you are an owner or admin and have opened dCal on the web in the last 30 days. With one, the assistant uses it without asking. With several, your app asks which one to use, and to change you just say "switch to" and the workspace's name. Behind the scenes, each call has to say which workspace it acts in:

* The assistant calls `list_my_accounts` to get the name and ID of each workspace, then sends the chosen ID as `account_id` with its calls.
* If a call does not say, dCal answers `ACCOUNT_REQUIRED` with the list of your workspaces, so the assistant can ask you and try again.
* The easiest fix is to tell the assistant up front which workspace to use, by its name.

Tools that work on a public booking page take the workspace from the link and never need `account_id`. The [MCP tools reference](https://docs.dapta.ai/dcal/developers/mcp-tools.md) has the details.

***

## What the connection can do

A sign-in acts with your owner or admin access, so it can use every tool on every event type of the workspace: find open times, hold a time, book, move a booking, confirm or decline requests, cancel, and read bookings, including what your guests wrote. It cannot edit event types, hours or settings, or manage webhooks, keys, members or teams. The **Tools** page in the left menu lists everything it can do, in plain words.

* **Your guests are emailed.** A booking, a move, a confirmation, a decline or a cancellation sends your guest the usual email, so a good assistant asks you before it acts. Cancelling and declining cannot be undone.
* **Some actions are marked destructive.** dCal marks moving, declining and cancelling a booking as destructive. Apps that honor that mark ask you before they run one.
* **Narrower access needs a key.** For an assistant that should only read, or only work with some event types, create an [API key](https://docs.dapta.ai/dcal/developers/api-keys.md) instead.
* **It keeps working while you stay an owner or admin.** If you go 30 days without opening dCal on the web in a workspace, the connection stops reaching that workspace until you open dCal there again. When it reaches no workspace at all, the assistant's calls answer `NOT_LINKED`. You do not need to reconnect.

***

## Disconnect

* **Your own connection:** remove the dCal connector in the app. Signed-in apps are not listed in dCal, so you always disconnect from the app.
* **Someone else's connection:** an owner can cut it off at once by removing that person's admin role, or removing them from the workspace, in **Settings** > **Members**. The assistant's next call is refused. See [Members and roles](https://docs.dapta.ai/dcal/account/members-and-roles.md).

***

## Troubleshooting

These messages come from dCal in English.

| What you see                                                                                                      | What to do                                                                                                                                                                                                                                                                                                                                                              |
| ----------------------------------------------------------------------------------------------------------------- | ----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- |
| The assistant reports `NOT_LINKED` and `This sign-in is not an owner or admin of any account on this server yet.` | Open dCal on the web at [app.dcal.ai](https://app.dcal.ai) with the same Dapta account (sign in again if it is already open), then ask again. There is no need to reconnect: the next call works. If it still fails, ask an owner to check that you are an admin or an owner in **Settings** > **Members**. Meanwhile, the tools for public booking pages keep working. |
| The app asks you to sign in again, or reports `Invalid or expired access token.`                                  | Reconnect the connector and sign in again.                                                                                                                                                                                                                                                                                                                              |
| The assistant reports `ACCOUNT_REQUIRED`                                                                          | You belong to several workspaces. Tell the assistant which one to use.                                                                                                                                                                                                                                                                                                  |
| The assistant works in the wrong workspace                                                                        | Tell it the workspace by name, or ask it to call `list_my_accounts` first.                                                                                                                                                                                                                                                                                              |
| You signed in with another Dapta account                                                                          | Remove the connector, add it again and sign in with the account you use for dCal.                                                                                                                                                                                                                                                                                       |

## Related

* [Connect an AI assistant (MCP)](https://docs.dapta.ai/dcal/developers/mcp-server.md): the agent pages in dCal, connecting with an API key, and the agent skill.
* [MCP tools reference](https://docs.dapta.ai/dcal/developers/mcp-tools.md): every tool, its inputs and its errors.
* [API keys](https://docs.dapta.ai/dcal/developers/api-keys.md): narrower access for an assistant or your own systems.
* [Members and roles](https://docs.dapta.ai/dcal/account/members-and-roles.md): who is an owner or an admin.
* [Security and privacy](https://docs.dapta.ai/dcal/security-and-privacy.md): how your guests' data is protected.


---

# Agent Instructions
This documentation is published with GitBook. GitBook is the documentation platform designed so that both humans and AI agents can read, navigate, and reason over technical content effectively. Learn more at gitbook.com.

## Querying This Documentation
If you need additional information that is not directly available in this page, you can query the documentation by asking a question.

Perform an HTTP GET request on the following URL with the `ask` and `goal` query parameters:

```
GET https://docs.dapta.ai/dcal/developers/connect-with-sign-in.md?ask=<question>&goal=<user_goal>
```

`ask` is the immediate question: it should be specific, self-contained, and written in natural language.
`goal` is what the user is ultimately trying to achieve, the reason they need the answer. Sharing it helps GitBook give you a better, more relevant answer. A goal is most helpful when it describes the outcome the user wants rather than restating the question. For example, with `ask=how do I create an API token`, a goal like `automate deployments from our CI pipeline` lets GitBook tailor the answer to that use case.

The response will contain a direct answer to the question and relevant excerpts and sources from the documentation.

Use this mechanism when the answer is not explicitly present in the current page, you need clarification or additional context, or you want to retrieve related documentation sections.
