Google Gemini

Configure Google Gemini models for Weaver

Weaver supports Google’s Gemini models through both API key authentication and Google Cloud Application Default Credentials (ADC).

Authentication Methods

API Key Authentication

The simplest method using a Gemini API key from Google AI Studio.

1. Get API Key

1. Visit Google AI Studio
2. Create a new API key
3. Copy the key

2. Configure Weaver

Option A: Configuration File

Add to ~/.weaver/config.json:

{
  "providers": {
    "gemini": {
      "api_key": "YOUR_GEMINI_API_KEY",
      "api_base": "https://generativelanguage.googleapis.com/v1beta"
    }
  },
  "agents": {
    "defaults": {
      "provider": "gemini",
      "model": "gemini-3-flash-preview"
    }
  }
}

Option B: Environment Variable

Add to .env:

GEMINI_API_KEY=YOUR_GEMINI_API_KEY

Google Cloud ADC Authentication

Use Google Cloud Application Default Credentials for automatic token management with Vertex AI.

1. Set Up GCP Credentials

# Install gcloud CLI
curl https://sdk.cloud.google.com | bash

# Initialize and authenticate
gcloud init
gcloud auth application-default login

2. Configure Weaver

Add to ~/.weaver/config.json:

{
  "providers": {
    "gemini": {
      "auth_method": "gcloud-adc",
      "api_base": "https://generativelanguage.googleapis.com/v1beta"
    }
  },
  "agents": {
    "defaults": {
      "provider": "gemini",
      "model": "gemini-3-flash-preview"
    }
  }
}

The ADC method automatically refreshes access tokens and handles GCP authentication. This is recommended for production deployments using Vertex AI.

Available Models

Weaver supports all Gemini model variants:

Model	Description	Context Window
gemini-3-flash-preview	Fast, efficient model (default)	128K tokens
gemini-2.0-flash-exp	Latest experimental model	1M tokens
gemini-1.5-pro	Most capable model	2M tokens
gemini-1.5-flash	Balanced speed and capability	1M tokens

Configuration Options

api_key · string

Gemini API key from Google AI Studio

api_base · string

Gemini API endpoint URL

auth_method · string

Authentication method: api_key (default) or gcloud-adc / adc

proxy · string

HTTP/HTTPS proxy URL for API requests (optional)

Model Parameters

Configure model behavior:

{
  "agents": {
    "defaults": {
      "model": "gemini-3-flash-preview",
      "max_tokens": 8192,
      "temperature": 0.7
    }
  }
}

max_tokens · integer · default: "8192"

Maximum tokens in response (uses max_completion_tokens for some models)

temperature · float · default: "0.7"

Controls randomness in responses (0.0 = deterministic, 2.0 = very random)

Usage Examples

Using Specific Models

Specify model in conversation:

# Use Flash model for speed
weaver chat --model gemini-3-flash-preview

# Use Pro model for complex tasks
weaver chat --model gemini-1.5-pro

With Proxy

Configure proxy for restricted networks:

{
  "providers": {
    "gemini": {
      "api_key": "YOUR_GEMINI_API_KEY",
      "proxy": "http://proxy.company.com:8080"
    }
  }
}

Implementation Details

Weaver’s Gemini integration uses:

• HTTPProvider for API key authentication with OpenAI-compatible format
• VertexAIProvider for GCP ADC authentication with automatic token refresh
• OpenAI-compatible /chat/completions endpoint format
• Automatic model namespace handling (strips google/ prefix)

Source: pkg/providers/http_provider.go:299-417, pkg/providers/vertex_provider.go

Troubleshooting

API Key Issues

# Verify API key is set
echo $GEMINI_API_KEY

# Test with curl
curl https://generativelanguage.googleapis.com/v1beta/models \
  -H "Authorization: Bearer $GEMINI_API_KEY"

GCP ADC Issues

# Check ADC credentials
gcloud auth application-default print-access-token

# Re-authenticate if needed
gcloud auth application-default login

# Verify cloud-platform scope
gcloud auth application-default print-access-token --scopes=https://www.googleapis.com/auth/cloud-platform

Common Errors

401 Unauthorized

• Verify your API key is correct
• For ADC: Run gcloud auth application-default login
• Check that the API key hasn’t expired

403 Permission Denied

• Ensure Vertex AI API is enabled in your GCP project
• Verify your account has the necessary permissions
• For ADC: Check you have cloud-platform scope

429 Rate Limit

• Check your API quota in Google AI Studio
• Implement exponential backoff in your application
• Consider upgrading to a paid tier

Next Steps

Provider Overview

Back to all providers

Agent Configuration

Configure agents with Gemini