Add example scripts and README for Gemini API usage with ADC authentication

2026-06-10 17:51:14 +02:00
parent 66ab4d9863
commit 3577a87948
4 changed files with 183 additions and 1 deletions
@@ -5,4 +5,7 @@ terraform.tfvars
 terraform.tfstate
 terraform.tfstate.backup

-tfplan*
+tfplan*
+
+# VS Code
+.vscode/*
@@ -0,0 +1,82 @@
+# Gemini API Python Examples (Gemini Enterprise Agent Platform (Vertex AI) / ADC Authentication)
+
+This directory contains examples for interacting with the Gemini API using the **`google-genai`** Python SDK and the **`gemini-3.5-flash`** model.
+
+To comply with enterprise security policies that disable static API keys, these examples use the **Gemini Enterprise Agent Platform** (Vertex AI) authenticated via secure **Application Default Credentials (ADC)**.
+
+## Core Features
+
+* **Gemini 3.5 Flash**: Offers frontier-level intelligence, advanced reasoning, and pro-level coding capability inside an ultra-fast, cost-efficient Flash tier.
+* **Agentic Capabilities**: Optimized out of the box for handling complex, multi-step agentic workflows and sustained reasoning tasks.
+* **Gemini Enterprise Agent Platform (Vertex AI) Integration**: Uses native Google Cloud IAM authorization instead of long-lived API keys.
+
+---
+
+## Prerequisites
+
+1. **Google Cloud Project**: You need an active Google Cloud Project.
+2. **Gemini Enterprise Agent Platform (Vertex AI) API Enabled**: Make sure the API (`aiplatform.googleapis.com`) is enabled in your project.
+3. **IAM Permissions**: Your Google account (or service account) must have the **Gemini Enterprise Agent Platform User** (Vertex AI User) (`roles/aiplatform.user`) role.
+4. **Google Cloud SDK (`gcloud`)**: Installed and configured on your machine.
+
+---
+
+## Setup Guide
+
+### 1. Install Dependencies
+Create a virtual environment (optional but recommended) and install the required dependencies:
+
+```bash
+# Create and activate a virtual environment
+python3 -m venv venv
+source venv/bin/activate
+
+# Install the modern google-genai library
+pip install -r requirements.txt
+```
+
+### 2. Authenticate locally with Application Default Credentials (ADC)
+Generate local credentials that the SDK will automatically detect:
+
+```bash
+gcloud auth application-default login
+```
+
+### 3. Set Environment Variables
+Specify your Google Cloud project and region:
+
+```bash
+# Set your GCP Project ID (Required)
+export GOOGLE_CLOUD_PROJECT="your-gcp-project-id"
+
+# Set your preferred location (Optional, defaults to us-central1)
+export GOOGLE_CLOUD_LOCATION="us-central1"
+```
+
+---
+
+## Running the Examples
+
+Run the script to verify the connection and see both standard content generation and config-based content generation in action:
+
+```bash
+python call_gemini.py
+```
+
+---
+
+## Troubleshooting
+
+### "Gemini Enterprise Agent Platform (Vertex AI) API has not been used in project..."
+Enable the API on your Google Cloud Console or use the `gcloud` CLI:
+```bash
+gcloud services enable aiplatform.googleapis.com --project="your-gcp-project-id"
+```
+
+### "Permission denied..." or "Lack of IAM permissions..."
+Ensure your authenticated user or active service account has the **Gemini Enterprise Agent Platform User** (Vertex AI User) role on your project:
+```bash
+gcloud projects add-iam-policy-binding your-gcp-project-id \
+  --member="user:your-email@example.com" \
+  --role="roles/aiplatform.user"
+```
@@ -0,0 +1,96 @@
+import os
+import sys
+from google import genai
+from google.genai import types
+
+def main():
+    # When API keys are disabled by organizational policy, we must authenticate using
+    # Google Cloud Application Default Credentials (ADC) via the Gemini Enterprise Agent Platform (Vertex AI) API.
+    
+    # Check for GCP project configuration
+    project_id = os.environ.get("GOOGLE_CLOUD_PROJECT")
+    location = os.environ.get("GOOGLE_CLOUD_LOCATION", "us-central1")
+    
+    if not project_id:
+        print("Error: GOOGLE_CLOUD_PROJECT environment variable is not set.", file=sys.stderr)
+        print("Please configure your GCP project using one of the following methods:", file=sys.stderr)
+        print("\nMethod 1: Set the environment variable directly:", file=sys.stderr)
+        print("  export GOOGLE_CLOUD_PROJECT=\"your-gcp-project-id\"", file=sys.stderr)
+        print("\nMethod 2: Authenticate using gcloud CLI Application Default Credentials (ADC):", file=sys.stderr)
+        print("  gcloud auth application-default login", file=sys.stderr)
+        print("  gcloud config set project your-gcp-project-id", file=sys.stderr)
+        sys.exit(1)
+
+    print("Initializing Gemini Client via Gemini Enterprise Agent Platform (Vertex AI) using Application Default Credentials (ADC)...")
+    print(f"Project ID: {project_id}")
+    print(f"Location:   {location}")
+    print("-" * 50)
+
+    try:
+        # Initialize the Client for Gemini Enterprise Agent Platform (Vertex AI) using Application Default Credentials (ADC).
+        # Setting vertexai=True routes the request through the Gemini Enterprise Agent Platform (Vertex AI Agent Engine).
+        client = genai.Client(
+            vertexai=True,
+            project=project_id,
+            location=location
+        )
+    except Exception as e:
+        print(f"Failed to initialize client: {e}", file=sys.stderr)
+        print("Please ensure you have run 'gcloud auth application-default login'", file=sys.stderr)
+        sys.exit(1)
+
+    # Use the gemini-3.5-flash model, which offers frontier-level reasoning,
+    # coding capability, and agentic workflows.
+    model_name = "gemini-2.5-flash"
+
+    # Prepare a simple text prompt
+    prompt = "Explain the concept of containerization in three simple sentences."
+
+    print(f"\nSending prompt to model '{model_name}':")
+    print(f"\"{prompt}\"")
+    print("-" * 50)
+
+    try:
+        # Send the prompt to Gemini Enterprise Agent Platform (Vertex AI) and get the response
+        response = client.models.generate_content(
+            model=model_name,
+            contents=prompt,
+        )
+
+        # Print the response text
+        print("Response received:")
+        print(response.text)
+        print("-" * 50)
+
+    except Exception as e:
+        print(f"An error occurred while generating content: {e}", file=sys.stderr)
+        print("\nTroubleshooting tips:", file=sys.stderr)
+        print("1. Ensure Gemini Enterprise Agent Platform (Vertex AI) API ('aiplatform.googleapis.com') is enabled in your project.", file=sys.stderr)
+        print("2. Ensure your authenticated identity has the \"Gemini Enterprise Agent Platform User\" (\"Vertex AI User\") role (roles/aiplatform.user).", file=sys.stderr)
+        sys.exit(1)
+
+    # Example 2: Sending a prompt with configuration options (temperature, etc.)
+    prompt_config = "Describe three advantages of microservices architecture."
+    print(f"\nSending prompt with configuration to model '{model_name}':")
+    print(f"\"{prompt_config}\"")
+    print("-" * 50)
+
+    try:
+        response_config = client.models.generate_content(
+            model=model_name,
+            contents=prompt_config,
+            config=types.GenerateContentConfig(
+                temperature=0.2,       # Lower temperature means more deterministic responses
+                max_output_tokens=300, # Limit the response length
+            )
+        )
+        print("Response received (with config):")
+        print(response_config.text)
+        print("-" * 50)
+
+    except Exception as e:
+        print(f"An error occurred: {e}", file=sys.stderr)
+        sys.exit(1)
+
+if __name__ == "__main__":
+    main()
@@ -0,0 +1 @@
+google-genai>=0.1.0