gcp-cli-ai 0.1.0

AI-powered CLI for executing GCP commands using natural language

GCP CLI

An AI-powered command-line interface for executing Google Cloud Platform commands using natural language.

Features

🤖 AI-Powered Command Generation - Use Vertex AI Gemini to generate GCP Python scripts from natural language queries

🛡️ Safe Execution - Preview generated code before execution with user confirmation

🔑 Flexible Authentication - Support for service account JSON files and Application Default Credentials

💻 Rich CLI Experience - Beautiful terminal output with syntax highlighting and interactive mode

📦 Modular Architecture - Easy to extend and customize for different GCP services

Installation

Prerequisites

• Python 3.8 or higher
• Google Cloud Platform account with Vertex AI enabled
• GCP credentials (service account JSON or Application Default Credentials)

Install from source

# Clone the repository
cd /Users/govind/Desktop/Govind/projects/Cloud_AI

# Install dependencies
pip install -r requirements.txt

# Install the package
pip install -e .

Quick Start

1. Set up credentials

Recommended: Use Application Default Credentials (ADC)

# Authenticate once - credentials stored securely
gcloud auth application-default login

That's it! The library will automatically use these credentials.

Alternative: Service Account Key (Not recommended for local development)

Only use this for CI/CD or production environments:

export GOOGLE_APPLICATION_CREDENTIALS="/path/to/your/credentials.json"

2. Run your first command

# Execute a simple query
gcp-cli execute "list all compute instances in us-central1"

# Start interactive mode
gcp-cli interactive

# Run a Python script
gcp-cli run my_script.py

Usage

Execute Mode

Execute a natural language GCP command:

gcp-cli execute "list all cloud storage buckets"
gcp-cli execute "create a compute instance named test-vm in us-central1-a"
gcp-cli execute "list all IAM service accounts"

Options:

• --no-preview: Skip code preview and execute immediately
• --dry-run: Generate code without executing
• --context: Add additional context for code generation

Interactive Mode

Start an interactive session for multiple commands:

gcp-cli interactive

In interactive mode, you can:

• Execute multiple commands in sequence
• View command history with history command
• Exit with exit or quit

Run Script

Execute an existing Python script:

gcp-cli run my_gcp_script.py

Configuration

Create a configuration file:

gcp-cli init-config --output myconfig.yaml

Example config file (myconfig.yaml):

project_id: my-project
location: us-central1
model: gemini-2.0-flash-exp
max_output_tokens: 8192
temperature: 1.0
top_p: 0.95
preview_before_execute: true
log_level: INFO

Note: Authentication uses Application Default Credentials (ADC).
Run gcloud auth application-default login before using the library.

Use a config file:

gcp-cli --config myconfig.yaml execute "list all instances"

View Information

Display current configuration:

gcp-cli info

View command history:

gcp-cli history

Python API

You can also use GCP CLI as a Python library:

from gcp_cli import GCPCommandExecutor, ConfigManager, CredentialManager

# Initialize
config = ConfigManager()
config.set('project_id', 'my-project')
config.set('location', 'us-central1')

credentials = CredentialManager(credentials_path='/path/to/creds.json')

executor = GCPCommandExecutor(config=config, credentials=credentials)

# Execute a query
result = executor.execute_natural_query(
    query="list all compute instances",
    preview=True,
    dry_run=False
)

print(result['output'])

Command Examples

Here are some example queries you can try:

Compute Engine

• "list all compute instances"
• "create a compute instance named my-vm in us-central1-a with machine type n1-standard-1"
• "stop the instance named my-vm"

Cloud Storage

• "list all storage buckets"
• "create a storage bucket named my-bucket in us-central1"
• "upload file data.txt to bucket my-bucket"

Cloud SQL

• "list all cloud sql instances"
• "create a postgres cloud sql instance named my-db"

IAM

• "list all service accounts"
• "create a service account named my-sa"
• "list IAM roles for project"

Configuration Options

Option	Description	Default
project_id	GCP project ID	None (auto-detected from credentials)
location	Default GCP location	us-central1
model	Vertex AI model to use	gemini-2.0-flash-exp
max_output_tokens	Max tokens for code generation	8192
temperature	Generation temperature	1.0
top_p	Generation top_p	0.95
preview_before_execute	Show preview before execution	true
log_level	Logging level	INFO

Environment Variables

You can override configuration with environment variables:

• GOOGLE_APPLICATION_CREDENTIALS: Path to credentials JSON
• GCP_CLI_PROJECT_ID: Project ID
• GCP_CLI_LOCATION: GCP location
• GCP_CLI_MODEL: Model name

Safety Features

• Code Preview: Review generated code before execution
• User Confirmation: Explicit confirmation required (unless --no-preview)
• Dry Run Mode: Test code generation without execution
• Command History: Track all executed commands
• Error Handling: Comprehensive error messages and logging
• Timeout Protection: 5-minute execution timeout

Project Structure

Cloud_AI/
├── gcp_cli/
│   ├── __init__.py          # Package initialization
│   ├── cli.py               # CLI interface
│   ├── executor.py          # Command execution
│   ├── ai_generator.py      # AI code generation
│   ├── credentials.py       # Credential management
│   ├── config.py            # Configuration handling
│   └── utils.py             # Utility functions
├── examples/
│   └── basic_usage.py       # Usage examples
├── setup.py                 # Package setup
├── requirements.txt         # Dependencies
└── README.md               # This file

Development

Running Tests

# Install development dependencies
pip install pytest pytest-cov

# Run tests
pytest tests/

Contributing

Contributions are welcome! Please ensure:

• Code follows PEP 8 style guidelines
• All tests pass
• New features include tests and documentation

Troubleshooting

"Failed to validate GCP credentials"

Authenticate with gcloud:

gcloud auth application-default login

"Project ID not found"

Provide project ID via:

• Config file: project_id: my-project
• Command line: --project my-project
• Environment: export GCP_CLI_PROJECT_ID=my-project

"Model not found"

Ensure Vertex AI is enabled in your project:

gcloud services enable aiplatform.googleapis.com

License

MIT License - see LICENSE file for details

Support

For issues and questions:

• Open an issue on GitHub
• Check the documentation
• Review existing issues for similar problems

Acknowledgments

Built with:

• Vertex AI - AI-powered code generation
• Click - CLI framework
• Rich - Terminal formatting
• Google Cloud Client Libraries