Development Agent with Gemini Enterprise Agent Platform

This example deploys a software development agent built with the Agent Development Kit to the Agent Runtime of Gemini Enterprise Agent Platform. The agent reads a feature spec, generates implementation code, and writes tests, all within a single managed session.

The agent definition stays portable, and only the deployment call is specific to the service. The managed runtime provides session isolation and memory across turns.

Requirements

• Python 3.10+
• A Google Cloud project with the Gemini Enterprise Agent Platform API (aiplatform.googleapis.com) enabled
• Application Default Credentials configured (gcloud auth application-default login)
• A Cloud Storage bucket used for staging

Steps for running this example in the shell

1. Install dependencies:

python -m venv .venv

# macOS/Linux:
source .venv/bin/activate

# Windows Command Prompt:
.venv\Scripts\activate.bat

# Windows PowerShell:
.venv\Scripts\Activate.ps1

pip install -r requirements.txt

2. Export the project settings as environment variables:

export GOOGLE_CLOUD_PROJECT="your-project-id"
export GOOGLE_CLOUD_LOCATION="us-central1"
export STAGING_BUCKET="gs://your-staging-bucket"

3. Run the script:

python dev_agent.py

Output

The script creates a remote agent in the managed runtime and streams the agent's reasoning and tool calls as it reads spec.md, generates the implementation, and writes the test file. Deployment takes several minutes the first time, and the resulting agent remains available until deleted.

Note: deploying to a managed runtime creates billable cloud resources. The spec file spec.md in this folder is a sample you can modify to test different feature descriptions.