Support Google Vertex AI with Application Default Credentials (ADC)

[kaovilai]

Feature Request

Add support for using Claude models via Google Cloud Vertex AI with Application Default Credentials (ADC), similar to how Claude Code supports Vertex AI.

Context

Many enterprise and GCP-native users access Claude models through Google Cloud Vertex AI rather than directly via the Anthropic API. Claude Code already supports this workflow natively by setting a few environment variables and authenticating via gcloud auth application-default login — no separate Anthropic API key needed.

The typical setup in Claude Code looks like:

export CLAUDE_CODE_USE_VERTEX=1
export CLOUD_ML_REGION="us-east5"
export ANTHROPIC_VERTEX_PROJECT_ID="your-gcp-project-id"
gcloud auth application-default login

Proposed Behavior

T3 Code could support a similar configuration where users can:

1. Select Vertex AI as a provider in settings (alongside direct Anthropic API, AWS Bedrock, etc.)
2. Authenticate via ADC — using gcloud auth application-default login, service account keys (GOOGLE_APPLICATION_CREDENTIALS), or workload identity
3. Configure GCP project and region for Vertex AI endpoint routing

This would allow organizations that route all AI traffic through GCP to use T3 Code without needing a separate Anthropic API key.

Related

• Issue [Bug]: Doesn't work with Claude models via LiteLLM #1225 covers a workaround using LiteLLM as a proxy to Vertex AI, but native ADC support would be a more seamless experience
• Claude Code's Vertex AI support: https://docs.anthropic.com/en/docs/claude-code/bedrock-vertex

Note

Responses generated with Claude

[kaovilai]

kaovilai#2 sample impl. once verified will open something here.

[kaovilai]