The review page lets you give feedback on an email the way you would on a document: you read the rendered email, select a sentence and comment on it. Your AI agent (Claude, ChatGPT, Codex, Cursor or any client connected to Lumail) reads your comments, edits the campaign and answers you in the same thread.

## Open the review page

- **From the editor:** open a campaign and click the chat bubble button next to **Focus mode** in the top bar.
- **From your agent:** after it creates or edits a campaign, your agent gives you the `reviewUrl` link.
- **By URL:** `https://lumail.io/orgs/<your-org>/campaigns/agent/<campaignId>`

The email is shown full screen and read-only, exactly as subscribers will see it.

## Leave feedback

1. Select any text in the email and click **Comment**.
2. Write what should change, then press **⌘↵** (or **Ctrl+Enter**).
3. Use **General** in the sidebar for feedback that is not about a specific passage.

Commented text stays highlighted. Click a highlight to open its thread. Each thread can be answered, resolved, reopened or deleted.

## Let your agent apply it

Click **Copy agent prompt** at the bottom of the sidebar and paste it into your agent. It asks the agent to:

1. read your open threads with `list_campaign_comments`,
2. apply each one with `edit_campaign`,
3. answer each thread with `reply_campaign_comment`, and resolve it once handled.

Keep the review page open: it refreshes on its own every couple of seconds, so you see the email change and the agent's replies appear without reloading. If the agent rewrites a passage you commented on, the thread shows **This text changed since the comment.**

## Tools

| Tool | What it does |
| --- | --- |
| `list_campaign_comments` | Returns the open threads (selected text, surrounding context, your comment and replies) and the `reviewUrl`. |
| `reply_campaign_comment` | Answers a thread as the AI agent, optionally resolving it. |
| `resolve_campaign_comment` | Resolves or reopens a thread without replying. |

All three need the `campaigns` permission on your API token or MCP connection. See the [tools reference](/docs/ai-integration/tools-reference) for the full schemas.