Claude Code setup

If you live in the terminal, build and manage your agent from a Claude Code session instead of clicking through the editor. The VerticalAI MCP gives Claude Code the same control plane the app uses, so your edits go live the same way they would from the editor.

Using a different assistant? Connect your assistant covers Claude Desktop, Cursor, VS Code, Windsurf, and ChatGPT as well.

Step 1: install the MCP

The MCP is a remote server you connect over HTTP. Add it once, scoped to your user so it is available in every project:

claude mcp add --transport http --scope user vai https://app.verticalai.com.au/api/mcp

The first time Claude Code reaches the server it asks you to sign in. Run the /mcp command, pick vai, and finish the browser consent screen. Claude Code stores the token and refreshes it for you, so you sign in once.

To confirm it worked, run /mcp again and check vai is listed with a tool count.

Step 2: build and manage your agent

Ask in plain language. Start by seeing what the MCP can reach:

List my agents.

Then build or change one:

Open my support agent. Tighten the greeting, add a tool that looks up an
order by its number, and tell me what's blocking publish.

Claude Code reads your workspace, makes the edits, and runs the same readiness checks the app does before you publish. You can switch between Claude Code and the app at any time; both edit the same agent, so nothing is lost.

What you can do next

Once vai is connected, ask Claude Code to:

• List and inspect agents across your workspace, and read an agent's full config in one call.
• Edit the agent in plain language: change the prompt, add or revise tools, set variables, adjust the voice, or tune behaviour.
• Check readiness before you publish, so you see what blocks go-live and what is an acknowledgeable risk.
• Review calls with the session and transcript tools when you are debugging how the agent behaved.

The capabilities track the in-app editor: anything the Builder Chat can do, you can do from your terminal, alongside workspace tasks like phone numbers and evals. For the editor walkthrough, see Agents.