Skip to content
Google Vertex AI logo

Google Vertex AI MCP server

Generate content and images using Gemini and Imagen models on Google Vertex AI.

Google Vertex AI is not a separate install. One Activepieces MCP server exposes it alongside 760+ other apps, behind a single URL.

Tools
3
Works with
Any MCP client
Auth
Custom auth
License
MIT, open source
Updated
October 2026

About the Google Vertex AI MCP server

Connect Google Vertex AI once in Activepieces and every action it ships becomes a tool your assistant can call, behind the same server URL as the rest of your apps. Your agent reaches Google Vertex AI and 760+ other apps over a single connection, on our cloud or a self-hosted install.

Tools

Google Vertex AI tools your AI can run

Once connected, your assistant can call any of these Google Vertex AI actions by asking in plain language.

  • Generate Content

    Call a Gemini model on Vertex AI to generate a text response.

    12 fields

  • Generate Image

    Generate an image from a text prompt using Google Imagen models on Vertex AI.

    4 fields

  • Custom API Call

    Make a custom API call to a specific endpoint

    10 fields

Give your assistant Google Vertex AI

Connect Google Vertex AI once and every tool above is callable from Claude, Cursor or any MCP client, behind the same server URL as 760+ other apps. Free to start.

Google Vertex AI MCP, answered.

No. That is the point of it. One Activepieces server exposes Google Vertex AI alongside every other piece you connect, so your client keeps a single connection.

Yes. Activepieces is open source and free to start, and the MCP server is included on every plan, from the free cloud tier to any self-hosted deployment.

Connect Google Vertex AI in Activepieces, enable the server under Settings → MCP Server, then paste your server URL into your client config. It authenticates over OAuth on first use.

It can call 3 tools, including Generate Content, Generate Image, Custom API Call, and chain them with steps in other apps inside the same run.

Connections use OAuth2 with limited scopes and are stored under 256-bit encryption, with no API to read them back, so an agent gets scoped access rather than your keys.