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

# Connect your AI tool

> Add the Condense MCP server to Claude, ChatGPT, Cursor, GitHub Copilot, Gemini CLI, Codex and other AI tools, and sign in with your Condense account.

Condense MCP is a remote server. Your AI tool reaches it over HTTP and signs you in with OAuth. You add one URL, sign in once in your browser, and the tools appear.

```text theme={null}
https://app.condense.io/api/mcp_servers/mcp
```

> [!NOTE]
>
> Condense MCP is in beta.

> [!NOTE]
>
> The server name is up to you. This page uses `condense`. Use the same name in the sign-in command your AI tool gives you. Menu names and steps vary between AI tools, and your tool's own documentation has the latest wording.

## Before you add the server

Install an AI tool that supports MCP, open it and sign in to it with your own account. You need that tool working before you add Condense to it.

## Add the server

Choose your AI tool from the list.

Add the server to Claude Code, for your project, from the command line:

```bash theme={null}
claude mcp add --scope project --transport http condense "https://app.condense.io/api/mcp_servers/mcp"
```

Or add this to `.mcp.json` in your project root:

```json .mcp.json theme={null}
{
  "mcpServers": {
    "condense": {
      "type": "http",
      "url": "https://app.condense.io/api/mcp_servers/mcp"
    }
  }
}
```

Then sign in. In a regular terminal (not the IDE extension), start Claude Code:

```bash theme={null}
claude
```

Type `/mcp` inside Claude Code, select the `condense` server and choose **Authenticate**. Your browser opens the Condense approval page.

Add the server to Codex from your terminal:

```bash theme={null}
codex mcp add condense --url "https://app.condense.io/api/mcp_servers/mcp"
```

Or add this to `~/.codex/config.toml`:

```toml config.toml theme={null}
[mcp_servers.condense]
url = "https://app.condense.io/api/mcp_servers/mcp"
```

Then sign in:

```bash theme={null}
codex mcp login condense
```

Add the server to Gemini CLI from your terminal:

```bash theme={null}
gemini mcp add --transport http condense "https://app.condense.io/api/mcp_servers/mcp"
```

Or add this to `~/.gemini/settings.json`, or to `.gemini/settings.json` in your project. In Gemini CLI the field is `httpUrl`:

```json settings.json theme={null}
{
  "mcpServers": {
    "condense": {
      "httpUrl": "https://app.condense.io/api/mcp_servers/mcp"
    }
  }
}
```

Start Gemini CLI and run:

```text theme={null}
/mcp auth condense
```

Add the server to GitHub Copilot CLI from your terminal:

```bash theme={null}
copilot mcp add --transport http condense-mcp https://app.condense.io/api/mcp_servers/mcp
```

Copilot CLI confirms the server it added.

![The terminal after adding the Condense MCP server to Copilot CLI](https://mintcdn.com/zeliot/otDRDVPickK4N8ah/images/mcp/copilot-cli-add-server.png?fit=max\&auto=format\&n=otDRDVPickK4N8ah\&q=85\&s=6cd53ec1ae1a53c85601687fd0f07d99)

Copilot CLI confirms the Condense MCP server.

Start the connection from Copilot CLI. Your browser opens the Condense approval page.

Add the server to Warp. Put this in `~/.warp/.mcp.json`:

```json .warp/.mcp.json theme={null}
{
  "mcpServers": {
    "condense": {
      "url": "https://app.condense.io/api/mcp_servers/mcp"
    }
  }
}
```

Start the connection from Warp. Your browser opens the Condense approval page.

In Claude Desktop or on claude.ai:

1. Open **Settings**, then **Connectors**.
2. Select **Add custom connector**.
3. Enter the name `Condense` and the URL `https://app.condense.io/api/mcp_servers/mcp`.
4. Select **Add**, then **Connect**, and sign in with your Condense account.

Add the server to ChatGPT. What you can add depends on your ChatGPT plan.

1. Open **Settings**, then **MCP servers**, and select **Add server**.
2. Name the server and choose **Streamable HTTP**.
3. Enter the URL `https://app.condense.io/api/mcp_servers/mcp`.
4. Save, then restart ChatGPT. Your browser opens the Condense approval page when you first use the server.

Add the server to Perplexity. Custom remote connectors depend on your Perplexity plan.

1. Open **Account settings**, then **Connectors**.
2. Select **Custom connector**, then **Remote**.
3. Enter the name `Condense`, the URL `https://app.condense.io/api/mcp_servers/mcp`, authentication **OAuth** and transport **Streamable HTTP**.
4. Acknowledge the risk notice and select **Add**.
5. Select the connector card to start the sign-in.

Add the server to Grok:

1. Go to `grok.com/connectors`.
2. Select **New Connector**, then **Custom**.
3. Enter the URL `https://app.condense.io/api/mcp_servers/mcp` and complete the sign-in.

Add the server to Cursor. Put this in `.cursor/mcp.json` in your project, or to `~/.cursor/mcp.json` to use it in every project:

```json .cursor/mcp.json theme={null}
{
  "mcpServers": {
    "condense": {
      "url": "https://app.condense.io/api/mcp_servers/mcp"
    }
  }
}
```

Open the **Customize** page from the Cursor sidebar, where your MCP servers are listed. Select the `condense` server and sign in when prompted.

Add the server to GitHub Copilot in VS Code. This adds it to your VS Code user profile, so it is available in every workspace:

```bash theme={null}
code --add-mcp '{"name":"condense","type":"http","url":"https://app.condense.io/api/mcp_servers/mcp"}'
```

Or add it to the current workspace only, in `.vscode/mcp.json`. VS Code uses `servers`, not `mcpServers`:

```json .vscode/mcp.json theme={null}
{
  "servers": {
    "condense": {
      "type": "http",
      "url": "https://app.condense.io/api/mcp_servers/mcp"
    }
  }
}
```

Open `.vscode/mcp.json` and select **Start** above the `condense` entry. VS Code asks you to sign in. Allow it and finish the sign-in in your browser. The Condense tools then appear in Copilot Chat in Agent mode.

Add the server to Windsurf. Put this in `~/.codeium/windsurf/mcp_config.json`. Windsurf uses `serverUrl` for a remote server:

```json mcp_config.json theme={null}
{
  "mcpServers": {
    "condense": {
      "serverUrl": "https://app.condense.io/api/mcp_servers/mcp"
    }
  }
}
```

Refresh the MCP servers list in Cascade and sign in when prompted. Do not add a `headers` entry, because Condense signs you in with OAuth.

Add the server to Google Antigravity. Put this in `~/.gemini/antigravity/mcp_config.json`. Antigravity needs `serverUrl`, and does not accept `url` or `httpUrl`:

```json mcp_config.json theme={null}
{
  "mcpServers": {
    "condense": {
      "serverUrl": "https://app.condense.io/api/mcp_servers/mcp"
    }
  }
}
```

If Condense does not load, check this field name first.

Add the server to Kiro. Put this in `~/.kiro/settings/mcp.json`:

```json mcp.json theme={null}
{
  "mcpServers": {
    "condense": {
      "url": "https://app.condense.io/api/mcp_servers/mcp"
    }
  }
}
```

Start the connection from Kiro. Your browser opens the Condense approval page.

Any AI tool that supports remote MCP servers and OAuth can connect. In your tool:

1. Find the option to add a new MCP server.
2. Enter the Condense MCP server URL in the server URL field.
3. Give the server a name, such as `Condense MCP`, if your tool asks for one.
4. Save the configuration.
5. Start the connection. Your browser opens the Condense approval page.

## Which field holds the URL

AI tools do not agree on how to write a remote server. Use the form below for your tool exactly as shown. The file each tool reads is named in its steps above.

| AI tool | Field for the URL | Top-level key |
| :- | :- | :- |
| Claude Code | `url`, with `"type": "http"` | `mcpServers` |
| OpenAI Codex CLI | `url` | `mcp_servers`, in a TOML file |
| Gemini CLI | `httpUrl` | `mcpServers` |
| Warp | `url` | `mcpServers` |
| Claude Desktop and claude.ai | Entered in the dialog | Not applicable |
| ChatGPT | Entered in the dialog | Not applicable |
| Perplexity | Entered in the dialog | Not applicable |
| Grok | Entered in the dialog | Not applicable |
| Cursor | `url` | `mcpServers` |
| GitHub Copilot (VS Code) | `url`, with `"type": "http"` | `servers` |
| Windsurf | `serverUrl` | `mcpServers` |
| Antigravity | `serverUrl` | `mcpServers` |
| Kiro | `url` | `mcpServers` |

## What happens next

The first sign-in opens a Condense page in your browser. There you choose the environment and the permissions your AI tool gets. Continue with [Sign in and approve access](https://docs.condense.io/condense/v3.0.0/mcp/sign-in-and-approve-access.md).

> [card] [Sign in and approve access](https://docs.condense.io/condense/v3.0.0/mcp/sign-in-and-approve-access.md)
>
> Choose the environment and the permissions.

> [card] [Verify and reconnect](https://docs.condense.io/condense/v3.0.0/mcp/verify-and-reconnect.md)
>
> Check the connection works, or sign in again.


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