Source: https://docs.pageloop.ai/mcp/connect

# Connect your AI tool

# Connect your AI tool

Connect your AI tool to Pageloop to work with your documentation from a conversation. Sign-in happens in your browser, where you choose what the tool can access.

## Before you start

- Use this server address: `https://mcp.pageloop.ai/mcp`.
- Have a Pageloop-hosted knowledge base: a docs site. Connected help centers are not supported yet.
- Check your Pageloop role. Your role sets what you can allow the tool to do; see [Choose what your AI tool can do](/mcp/choose-access#what-your-role-can-grant).

## Add Pageloop to your AI tool

### In a chat app (web or desktop)

#### Claude

1. Open **Settings → Connectors → Add custom connector**.
2. Name the connector **Pageloop** and paste `https://mcp.pageloop.ai/mcp` as the server address.
3. Select **Add**, then **Connect**.

On Team and Enterprise plans, an Owner adds the connector first under **Organization settings → Connectors**. Each member then connects the tool for their own account.

#### Other chat apps

1. In your app's settings, add a custom connector or remote MCP server.
2. Use `https://mcp.pageloop.ai/mcp` as the server address.
3. Sign in when prompted.

### In a terminal or code editor

#### Claude Code

1. Run this command in your terminal:

   ```bash
   claude mcp add --transport http pageloop https://mcp.pageloop.ai/mcp
   ```

2. In Claude Code, enter `/mcp`, choose **pageloop**, then select **Authenticate** to open browser sign-in.

#### Codex

1. Add Pageloop:

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

2. Start sign-in:

   ```bash
   codex mcp login pageloop
   ```

#### Gemini CLI

1. In `~/.gemini/settings.json`, add this entry under `mcpServers`, keeping any servers you already have:

   ```json
   "pageloop": {"httpUrl": "https://mcp.pageloop.ai/mcp"}
   ```

2. In Gemini CLI, run `/mcp auth pageloop` to start sign-in.

#### Cursor, VS Code and other editors

1. Open your editor's MCP settings and add a remote **HTTP** MCP server named **pageloop**.
2. Use `https://mcp.pageloop.ai/mcp` as the server address.
3. Sign in when prompted.

## Sign in and choose access

1. Sign in to Pageloop if prompted. Check the account shown under **Signed in as** and the name of the tool requesting access.
2. Choose the **Workspace** you want the tool to use.
3. Under **Knowledge bases**, tick the docs sites the tool should access. Tick at least one.
4. Under **Access**, choose what the tool may do. **Propose changes** is marked **Recommended** for drafting and reviewing without publishing live. Read [Choose what your AI tool can do](/mcp/choose-access) before granting broader access.
5. Select **Allow** to approve the connection. For **Read only**, the button says **Allow read access**; for **Full access**, it says **Allow full access**.
6. Return to your AI tool to finish connecting.

![Pageloop sign-in screen asking to connect an AI tool, before a workspace is chosen](/images/uploads/01-consent-start-e6faf5f6.png)

![Workspace list open on the Pageloop sign-in screen](/images/uploads/02-consent-workspace-picker-e6093ec9.png)

### “Self-registered application”

**Self-registered application** means the tool registered itself with Pageloop. This is normal for MCP tools. Check that the name shown matches the tool you are connecting before you allow access. If you did not start the connection or the name is not what you expect, select **Cancel** instead.

## Check that it works

Ask your tool: “List my documentation pages.” The result should show pages from the knowledge bases you selected.

Your tool's documentation edits stay in a change set for review unless you grant publishing access. You can review the connection or end its access in [Manage and disconnect AI tools](/mcp/manage-connections).

If sign-in fails or an expected capability is missing, see [Pageloop MCP troubleshooting](/guides/troubleshooting#pageloop-mcp).
