# Connect your AI

> Add Dockito to Claude Code, Claude, ChatGPT, Cursor, VS Code and any MCP client.

Connect once and your AI can publish docs, read them and keep their History, through the Dockito MCP server. It can never do more than you can, and only what you allow when you connect it.

Pick your tool:

### Claude Code

**Command**

1. Add Dockito:

   ```sh
   claude mcp add --transport http dockito https://dockito.ai/mcp
   ```
2. In Claude Code, run `/mcp`, choose **dockito**, then **Authenticate**.
3. Your browser opens Dockito. Choose what it may do in each workspace (**View**, **Publish** or **Everything**) and **Approve**.

**Token**

1. In Dockito, open **Account → Tokens** and make a token, with **Publish** where it should write.
2. Add Dockito with it:

   ```sh
   claude mcp add --transport http dockito https://dockito.ai/mcp \
     --header "Authorization: Bearer dkt_…"
   ```

Add `--scope user` to have Dockito in every project.

### Claude

1. Open **Settings → Connectors** and choose **Add custom connector**.
2. Name it **Dockito**, paste the address and choose **Connect**:

   ```text
   https://dockito.ai/mcp
   ```
3. Your browser opens Dockito. Choose what it may do in each workspace (**View**, **Publish** or **Everything**) and **Approve**.
4. In a chat, turn Dockito on from the tools menu.

Custom connectors need a paid Claude plan. On Team and Enterprise, an owner may need to add it.

### ChatGPT

1. Open **Settings → Apps & Connectors → Advanced settings** and turn on **Developer mode**.
2. Back in **Apps & Connectors**, choose **Create**. Name it **Dockito**, paste the address and pick **OAuth**:

   ```text
   https://dockito.ai/mcp
   ```
3. Your browser opens Dockito. Choose what it may do in each workspace (**View**, **Publish** or **Everything**) and **Approve**.
4. In a chat, add Dockito from the **+** menu under **Developer mode**.

Developer mode is on paid ChatGPT plans. On Business and Enterprise, an admin may need to allow it.

### Cursor

**One click**

[Add to Cursor](cursor://anysphere.cursor-deeplink/mcp/install?name=dockito&config=eyJ1cmwiOiJodHRwczovL2RvY2tpdG8uYWkvbWNwIn0=)

1. Choose **Add to Cursor** and confirm in Cursor.
2. In **Cursor Settings → MCP**, choose **Needs login** next to dockito.
3. Your browser opens Dockito. Choose what it may do in each workspace (**View**, **Publish** or **Everything**) and **Approve**.

**Config file**

1. Add Dockito to `~/.cursor/mcp.json`:

   ```json
   {
     "mcpServers": {
       "dockito": {
         "url": "https://dockito.ai/mcp"
       }
     }
   }
   ```
2. In **Cursor Settings → MCP**, choose **Needs login** next to dockito.
3. Your browser opens Dockito. Choose what it may do in each workspace (**View**, **Publish** or **Everything**) and **Approve**.

**Token**

1. In Dockito, open **Account → Tokens** and make a token, with **Publish** where it should write.
2. Add Dockito to `~/.cursor/mcp.json` with it:

   ```json
   {
     "mcpServers": {
       "dockito": {
         "url": "https://dockito.ai/mcp",
         "headers": {
           "Authorization": "Bearer dkt_…"
         }
       }
     }
   }
   ```

### VS Code

**One click**

[Add to VS Code](vscode:mcp/install?%7B%22name%22%3A%22dockito%22%2C%22type%22%3A%22http%22%2C%22url%22%3A%22https%3A%2F%2Fdockito.ai%2Fmcp%22%7D)

1. Choose **Add to VS Code** and confirm **Install** in VS Code.
2. Allow it to sign in to Dockito when it asks.
3. Your browser opens Dockito. Choose what it may do in each workspace (**View**, **Publish** or **Everything**) and **Approve**.

**Config file**

1. Add Dockito to `.vscode/mcp.json` in your project:

   ```json
   {
     "servers": {
       "dockito": {
         "type": "http",
         "url": "https://dockito.ai/mcp"
       }
     }
   }
   ```
2. Choose **Start** above the server, and allow it to sign in.
3. Your browser opens Dockito. Choose what it may do in each workspace (**View**, **Publish** or **Everything**) and **Approve**.

Use Dockito from Copilot Chat in **Agent** mode.

### Antigravity

1. In Dockito, open **Account → Tokens** and make a token, with **Publish** where it should write.
2. In the agent panel, open the **⋯** menu → **MCP Servers → Manage MCP Servers → View raw config**.
3. Add Dockito:

   ```json
   {
     "mcpServers": {
       "dockito": {
         "serverUrl": "https://dockito.ai/mcp",
         "headers": {
           "Authorization": "Bearer dkt_…"
         }
       }
     }
   }
   ```
4. Save, then choose **Refresh**.

### Codex

1. Add Dockito to `~/.codex/config.toml`:

   ```toml
   [mcp_servers.dockito]
   url = "https://dockito.ai/mcp"
   ```
2. Sign in:

   ```sh
   codex mcp login dockito
   ```
3. Your browser opens Dockito. Choose what it may do in each workspace (**View**, **Publish** or **Everything**) and **Approve**.

### Gemini CLI

1. Add Dockito to `~/.gemini/settings.json`:

   ```json
   {
     "mcpServers": {
       "dockito": {
         "httpUrl": "https://dockito.ai/mcp"
       }
     }
   }
   ```
2. In Gemini CLI, run `/mcp auth dockito`.
3. Your browser opens Dockito. Choose what it may do in each workspace (**View**, **Publish** or **Everything**) and **Approve**.

### Any MCP client

**Sign in**

1. Add a remote (HTTP) MCP server with this address:

   ```text
   https://dockito.ai/mcp
   ```
2. If your client signs in with OAuth, it opens Dockito for you.
3. Your browser opens Dockito. Choose what it may do in each workspace (**View**, **Publish** or **Everything**) and **Approve**.

**Token**

1. In Dockito, open **Account → Tokens** and make a token, with **Publish** where it should write.
2. Add the address, and send the token as a header:

   ```text
   https://dockito.ai/mcp
   Authorization: Bearer dkt_…
   ```

## What to ask

Your AI reads Dockito's instructions when it connects, so plain words work:

- “Publish this spec to Dockito and give me the link.”
- “Put these notes in the Harbor project, in Meetings.”
- “What changed in the Cedarline brief since Friday? Restore the version from before.”
- “Save the current version of the proposal as *Sent to the client*.”

It's told to **ask where a doc goes** rather than guess, and to **keep docs private** unless you ask for a public link. Deleting docs and sharing with people happen in the Dockito app, never through an AI.

## More

- [Tokens and access](/help/agents/tokens): what a connection may do, and how to sign it out.
- [MCP tools](/help/agents/mcp-tools): every tool and its parameters.