kojo-deploy-ai

The AI Orchestrator for Google Cloud Run.

kojo-deploy-ai is a Gemini CLI extension that brings natural language power to your Google Cloud infrastructure. Built on the Model Context Protocol (MCP), it wraps the kojo-deploy toolkit to allow you to manage, monitor, and push containerized services to Google Cloud Run using simple conversational commands.

Why kojo-deploy-ai?

• Zero-Install Execution: Uses npx to ensure you always have the latest deployment logic without global bloat.
• AI-Driven Workflows: Instead of memorizing complex flags, tell Gemini what you want to achieve.
• GCP Optimized: Specifically tuned for Google Cloud Run environments, handling service names, regions, and environment flags natively.
• MCP Standard: Built on the Model Context Protocol, ensuring a standardized "bridge" between the LLM and your local system.
• Windows Optimized: Includes native support for Windows environments, resolving common execution policy and path issues.

Prerequisites

• Gemini CLI: Install the official AI agent for your terminal: npm install -g @google/gemini-cli
• Google Cloud CLI: Ensure the gcloud tool is installed and authenticated (gcloud auth login) so the extension can interact with your Google Cloud account.
• Node.js: Version 20 or higher is required.

Installation

You can install kojo-deploy-ai directly from GitHub with one command:

gemini extensions install https://github.com/KojoShaddy/kojo-deploy-ai

Getting Started

Once installed, simply type gemini in your terminal to start the interactive mode. You don't need to manually create any configuration files. Just ask the AI to help you get started:

"I want to deploy this folder to Google Cloud Run. Can you help me set it up?"

Usage Examples

Open the Gemini CLI by typing gemini in your terminal. You can now manage your Cloud Run services using natural language:

Deploying to Cloud Run

"Deploy the billing-api service to production."

• Action: Executes npx kojo-deploy push billing-api --env prod

Checking Service Status

"What is the current status of my web-frontend on Google Cloud?"

• Action: Executes npx kojo-deploy status web-frontend

Intelligent Troubleshooting

"The last deployment failed. Can you check the status of the auth-service and explain what might be wrong?"

How it Works

This extension acts as an MCP Server. It creates a virtual bridge using Standard I/O to communicate with the Gemini CLI. When you give a command, the AI maps your intent to specialized tools:

Tool	Action	Description
initialize_project	npx kojo-deploy init	Automatically creates kojo.json with your project settings.
deploy_service	npx kojo-deploy push	Triggers a Cloud Build and deploys a new revision to Cloud Run.
get_deploy_status	npx kojo-deploy status	Fetches the URL, traffic split, and health of the service.

Troubleshooting

Windows Installation Issues

If you encounter an EBUSY error or folder locking issues while installing on Windows, please refer to the Windows Fix Guide.

Uninstallation & Disabling

1. The Standard Uninstall

The cleanest way to remove it completely from your Gemini CLI is by using the uninstall command:

gemini extensions uninstall kojo-deploy-ai

2. The "Soft" Disable

If you don't want to delete it but just want to turn it off so Gemini stops using those tools for a while, you can simply disable it:

gemini extensions disable kojo-deploy-ai

To turn it back on later, just run gemini extensions enable kojo-deploy-ai.

Local Development

If you want to modify the extension or test changes locally:

1. Clone the repository.
2. Install dependencies: npm install.
3. Link the extension to your Gemini CLI:

   gemini extensions install .

Contributing

Whether it's adding support for Cloud Functions, Secret Manager, or improving the AI's reasoning, feel free to open an issue or submit a PR.

License

This project is licensed under the MIT License.

---

Built with ❤️ by Shadrack Inusah (Kojo Shaddy) Software Developer | Open Source Contributor | Lead @ GDG Accra