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

# Connect an AI client

> Use Pretty chats from Codex, Claude Code, or another MCP client.

Connect Codex, Claude Code, or another MCP client to Pretty. You need a Pretty account in a workspace with external access enabled. See [Pretty MCP](/mcp/overview) for what the connection can do.

The server URL is:

```text theme={null}
https://api.make-pretty.com/mcp
```

If authorization says your workspace cannot be reached from outside the app, ask your workspace admin to [contact support](/help/support).

## Connect with Codex

Add Pretty as a remote MCP server:

```bash theme={null}
codex mcp add pretty --url https://api.make-pretty.com/mcp
codex mcp login pretty
```

Sign in to Pretty in the browser window that opens. Check the workspace named on the consent screen, then allow access. The connection uses that workspace; switch to the workspace you want in the [web editor](https://editor.make-pretty.com) before authorizing if you belong to more than one.

Run `codex mcp list` to check that Pretty is configured. Start a Codex session and ask it to use Pretty's `list_chats` tool to verify access. The Codex CLI and IDE extension share MCP configuration. See the [Codex MCP guide](https://developers.openai.com/codex/extend/mcp) for client setup details.

## Connect with Claude Code

```bash theme={null}
claude mcp add --transport http pretty https://api.make-pretty.com/mcp
```

In Claude Code, run `/mcp`, select Pretty, choose **Authenticate**, and complete the browser sign-in and consent flow. Run `claude mcp list` to check that Pretty shows as connected. The command adds Pretty to the current project. To make it available in every project, add `--scope user` to the `claude mcp add` command. See the [Claude Code MCP guide](https://code.claude.com/docs/en/mcp-quickstart) for client setup details.

## Connect another MCP client

Choose **Streamable HTTP** as the transport, enter the server URL above, and use the client's OAuth sign-in. Pretty publishes its authorization details at [`/.well-known/oauth-protected-resource/mcp`](https://api.make-pretty.com/.well-known/oauth-protected-resource/mcp). The client must request the `mcp:tools` scope.

## Try your first chat

Ask your AI client:

```text theme={null}
Use Pretty to start a chat: draft a three-slide presentation about our quarterly results.
Tell me when the chat finishes and show me Pretty's reply.
```

The client calls `start_chat` and receives a `chatId`. It can call `get_chat_status` until the status is `finished`, `failed`, or `waiting_for_you`. For a question from Pretty, answer in the same chat with `send_message`. Use `list_messages` to read more than the latest reply.

To work from a file already in your Pretty Workspace, ask the client to find its path with `list_files` and attach that path to `start_chat` or `send_message`. You can also attach a URL. MCP clients cannot upload a new local file through this connection.

See the [MCP tool reference](/mcp/tools) for inputs, results, and chat statuses.