VerticalAI docs

Connect your assistant

Connect the VerticalAI MCP to Claude Code, Claude Desktop, Cursor, VS Code, Windsurf, or ChatGPT and build and manage your voice agent in plain language.

VerticalAI runs a remote MCP server that gives an AI assistant the same control plane the app uses. Connect it once and you can build, inspect, and manage your voice agent in plain language from whatever client you already work in: Claude Code, Claude Desktop, Cursor, VS Code, Windsurf, or ChatGPT. Edits go live the same way they would from the editor, so you can move between your assistant and the app and never lose work.

The server lives at:

https://app.verticalai.com.au/api/mcp

There are no keys to copy. The first time a client reaches the server it opens a browser to sign you in. Access is scoped to your workspace and you can revoke it at any time.

Paste-into-any-client config

Many MCP clients accept a JSON config. Use this exact block:

{
  "mcpServers": {
    "vertical-ai": { "type": "http", "url": "https://app.verticalai.com.au/api/mcp" }
  }
}

Claude Code

Add the server 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

Then run /mcp, pick vai, and finish the browser consent screen. Claude Code stores the token and refreshes it, so you sign in once. Run /mcp again to confirm vai is listed with a tool count.

For the full Claude Code walkthrough, see Claude Code setup.

Claude Desktop, Cursor, VS Code, Windsurf

These clients connect via the JSON config block above. Open the client's MCP servers settings, paste the block, and restart the client:

  • Claude Desktop: Settings → Developer → Edit Config, paste into mcpServers, save, and restart Claude Desktop.
  • Cursor: Settings → MCP (or ~/.cursor/mcp.json), add the vertical-ai entry, and reload.
  • VS Code: open the MCP servers configuration (mcp.json) and add the vertical-ai entry; restart the MCP server from the command palette.
  • Windsurf: Settings → Cascade → MCP servers, add the entry, and refresh.

The first time the client calls a tool it opens a browser to sign you in. After that the connection stays authorised.

ChatGPT

ChatGPT can connect to a remote MCP server as a custom connector:

  1. Open Settings → Apps → Advanced settings and turn on Developer mode.
  2. Choose Create app and add the remote MCP server URL https://app.verticalai.com.au/api/mcp.
  3. Sign in through the browser when prompted.

Plus and Pro get read-only connectors

ChatGPT gates write-capable custom connectors to Business, Enterprise, and Edu workspaces. On Plus / Pro individual plans, custom connectors are read-only, so you can list and inspect your agent but not edit it. Full management (changing the prompt, adding tools, publishing) needs a Business, Enterprise, or Edu seat. See OpenAI Help, "Developer mode and full MCP connectors in ChatGPT".

What you can do once connected

Ask in plain language. Start by seeing what the assistant 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.

The assistant reads your workspace, makes the edits, and runs the same readiness checks the app does before you publish. Anything the Builder Chat can do, you can do from your assistant: list and inspect agents, edit the prompt, add or revise tools, set variables, adjust the voice, check readiness, and review calls. For the editor walkthrough, see Agents.

On this page