Claude Code

If you already write in Claude Code, you do not have to move the message into unpitch to test it. Connect the two and Claude Code can reach your workspace directly: it pulls the context you have written down, saves the draft as a document, and runs the same Quality Check and Simulation you would run on the website.

The reading still happens in unpitch. Claude Code does the writing; unpitch supplies what it knows about your business and tells you how the message lands.

Setting it up

Setup is three steps and lives on its own page: Set up Claude Code. You install the connection, authorise one workspace in the browser, and confirm it worked. There is no API key to paste and no secret to keep.

What changes once it is connected

Nothing about how unpitch works changes. The documents Claude Code creates are ordinary documents — they appear in Recents, they open in the editor, they keep their history. A Quality Check started from Claude Code is the same run you would have started from the rail, and it draws on the same included allowance.

What you gain is that the loop closes without a copy-paste: write, check, read the findings, rewrite, all in one place.

What it is not

It is not a second way to buy anything, and it is not a sending tool. Claude Code cannot launch a campaign, read replies, or move a message anywhere. It also never edits your Knowledge Base — see Permissions and your workspace.

Claude Code is the only client unpitch supports today. The connection uses the standard Model Context Protocol, so other clients may be able to speak to it, but nothing else is tested or supported.