> ## 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.

# Verify and reconnect

> Check that your AI tool can reach Condense, and sign in again if the connection expires or is disconnected.

## Verify the connection

After you accept on the approval page, check that your AI tool can talk to Condense.

**1. Open your AI tool's chat**

Use the chat or prompt window of the AI tool you connected.

**2. Ask for your workspaces**

```text theme={null}
List all Condense workspaces I can access.
```

**3. Check the result**

Your AI tool calls the `get_all_workspaces` tool and shows the workspaces you can access.

A working connection looks like this:

* Your AI tool calls the Condense tool successfully.
* The workspaces you can access are listed.
* Workspaces outside your authorized access are not shown.

![Copilot CLI calling get\_all\_workspaces and showing a table of workspace IDs and names](https://mintcdn.com/zeliot/otDRDVPickK4N8ah/images/mcp/verify-list-workspaces.png?fit=max\&auto=format\&n=otDRDVPickK4N8ah\&q=85\&s=ed85dde4dfd08842c2f81c577ae20349)

Copilot CLI listing the workspaces it can reach through Condense MCP.

> [!NOTE]
>
> Some AI tools load a server's tools only when they need them. You may see your tool search for `get_all_workspaces` before it calls it. This is normal.

### If no tools appear

* **The sign-in did not finish.** Run your AI tool's sign-in step again. In Claude Code, type `/mcp`, select `condense` and choose **Authenticate**. In Gemini CLI, run `/mcp auth condense`. In Codex, run `codex mcp login condense`.
* **The URL or the field name is wrong for your AI tool.** Check the URL and the setting for your tool in [Connect your AI tool](https://docs.condense.io/condense/v3.0.0/mcp/connect-your-ai-tool.md#which-field-holds-the-url).
* **No environment is assigned to your account.** The approval page showed **No Environments Assigned**. Ask your administrator for access.
* **The tool is switched off for you.** You see **Access denied**. Ask your Condense administrator to enable it.

## Reconnect or sign in again

If the connection expires, is disconnected, or needs authorizing again:

**1. Open your AI tool's MCP settings**

Go to the MCP server or integration settings of your AI tool.

**2. Find the Condense server**

Locate the Condense MCP server you configured.

**3. Reconnect**

Select the reconnect or authenticate option.

**4. Sign in again**

Complete the Condense sign-in. You see the approval page again.

**5. Verify**

Check that the server shows as connected, then ask for your workspaces as above.

How you start the sign-in depends on your AI tool.

* **Claude Code:** run `/mcp`, select `condense` and choose **Authenticate**.
* **Claude Desktop and claude.ai:** open **Settings**, then **Connectors**, and select **Connect** on the Condense connector.
* **Cursor:** open **Customize**, select the `condense` server and sign in.
* **GitHub Copilot (VS Code):** select **Start** above the `condense` entry in `.vscode/mcp.json` and sign in.
* **Gemini CLI:** run `/mcp auth condense`.
* **OpenAI Codex CLI:** run `codex mcp login condense`.

> [!WARNING]
>
> If an administrator revoked your access or paused the connection, signing in again does not help until your access is restored. See [Errors and troubleshooting](https://docs.condense.io/condense/v3.0.0/mcp/errors-and-troubleshooting.md).


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