Claude Code /setup-vertex — Set Up Your Google Vertex AI Connection

Sometimes your company uses Google Cloud (GCP), or security policy requires calling models through GCP. /setup-vertex is the setup command that connects Claude Code to Google Vertex AI in these cases.

Definition (what it is)

/setup-vertex is a command that configures the connection to Google Vertex AI (GCP's AI model operation platform). It applies only to users in this environment.

Vertex AI is a service that lets you call various AI models from Google Cloud. If your company uses a GCP environment, you can have Claude Code operate through Vertex instead of calling models directly. Because it requires credentials (cloud access permission information) and project (the unit that groups resources in GCP) settings, it falls under the advanced category.

How to use it (by level)

Basic — Configure the Vertex connection

Type this in the input box.

/setup-vertex

This starts the procedure for setting Vertex authentication and project. Follow the guidance and specify your GCP credentials and the project to use, and the connection is configured. Credentials differ per person, so enter your own values issued by your company in place. For example, you fill in the fields like this.

GCP Project ID: <project_ID_to_use>
Region: <region_to_use_e.g._asia-northeast3>
Service Account Key Path: <issued_key_file_path>

Common pitfalls

• This command applies only to users in a Vertex environment. If you're using a personal account for general use, there's no need to set it up.
• Credentials and the service account key are sensitive information like a password. The <...> parts in the example above are just placeholders; the actual values must be handled securely per your company's security procedures, and be careful not to expose them externally.
• Entering the wrong project ID and region prevents connection. Verify the values your company designated and enter them accurately.

Real-world example

It fits well for teams that must process data only within GCP and whose infrastructure is all on Google Cloud. Connect to fit the environment with /setup-vertex, and you can use Claude Code as-is while honoring your company's infrastructure rules.

Taking it further

If you're on an AWS environment, there's a separate /setup-bedrock that works the same way. Viewing it alongside /config, which handles overall settings, is great for checking and tidying the connection state that fits your team's environment. If you get the project ID and credentials from your cloud administrator in advance before setup, you can finish in one go.

Wrap-up

/setup-vertex is the setup command that connects Claude Code to Google Vertex AI (for users in that environment only). Match authentication and project to your GCP infrastructure rules, and you can keep working as-is while honoring policy. The key is to use your own values, not placeholders, for credentials, and to handle them so they aren't exposed.

Based on Claude Code v2.1.154 (2026.05)