MCP server
Hookpair inside Claude and Cursor.
Ask your assistant to mine objections from reviews, look up competitors' running Meta ads, or prepare a message-match audit. It calls Hookpair through the Model Context Protocol.
Ask in plain words.
Your assistant picks the right Hookpair tool. A few things to try:
Claude, with Hookpair connected
Tap a prompt to copy it.
Server URL
https://hookpair.com/api/mcp
Streamable HTTP, no sign-in needed. Limited to 120 requests per 15 minutes per network.
Connect
- Claude (web and desktop)Settings → Connectors → Add custom connector, then paste the URL above.
- Claude CodeRun
claude mcp add --transport http hookpair <URL> - CursorAdd it to
~/.cursor/mcp.json(all projects) or.cursor/mcp.json(one project):{ "mcpServers": { "hookpair": { "url": "<URL>" } } }
What your assistant can do
- audit_message_match
- Prepares a Hookpair message-match audit of an ad and its landing page. Returns a link that opens the audit form in the user's Hookpair workspace with the page filled in; the audit itself (score, 7 continuity dimensions, quoted gaps, page draft) runs there. No score is returned by this tool.
- extract_customer_objections
- Finds recurring objections, pains and praise in customer reviews or comments (English and German keywords) and returns verbatim quotes plus FAQ prompts to answer them. Rule-based and deterministic; the text is processed for this call only and not stored.
- search_competitor_ads
- Searches the Meta Ad Library for active ads matching a brand or keyword and sorts them by how long they have been running. Requires the user's own Meta access token with Ad Library API access; it is used for this request only and not stored.
What it doesn't do
It doesn't run full audits inside the chat: scores, quoted gaps and page drafts are made in your Hookpair workspace, and the audit tool hands the page over with a link. Ad Library searches need your own Meta access token; it's used for that request only and not stored. Review text you send is analysed for that call and not kept.
Prefer a page? The same objection mining runs in the browser in the Objection Extractor, with more free tools next to it.